# Gemini 3 NYC Hackathon: Project Gallery

- **Event:** [Gemini 3 NYC Hackathon](https://cerebralvalley.ai/e/gemini-3-nyc-hackathon)
- **When:** Sat, Feb 28 at 9:00 AM – 9:00 PM (EST)
- **Where:** The Malin Flatiron, New York, NY
- **Hosts:** [Google DeepMind](https://cerebralvalley.ai/u/deepmind), [Cerebral Valley](https://cerebralvalley.ai/u/cv)
- **Projects:** 58 (6 placed)
- **Page:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery

## Projects

### 1. Score!

We are Score a crazy ai engine for crowds and narration!

https://clever-gratitude-production.up.railway.app/

- **Team:** [Roberto Banbanaste](https://cerebralvalley.ai/u/rdbhere), [Elie Nabushosi](https://cerebralvalley.ai/u/Elie)
- **GitHub:** https://github.com/Banbanaste/score
- **Demo video:** https://drive.google.com/file/d/12YlKW5_DthAN17Wio_rrTvPvGQfLPYwj/view?usp=sharing
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=45

### 2. Syncro

Syncro is an AI-powered editor that keeps your narration and footage aligned. Drop in clips and a script or record voiceover and Syncro helps you order clips, match them to the script, generate TTS and style captions. One timeline, one flow.

- **Placement:** Finalist
- **Team:** [Grishma Deshmukh](https://cerebralvalley.ai/u/gish), [Anushka Vijay Kumar Naik](https://cerebralvalley.ai/u/anushka_naik), [Vishvesh Trivedi](https://cerebralvalley.ai/u/nerdyvisky), [Om Kamath](https://cerebralvalley.ai/u/omkamath)
- **GitHub:** https://github.com/Grishmadeshmukh/Syncro
- **Demo video:** https://www.youtube.com/watch?v=-2adTCH8aMs
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=48

### 3. TasteTale

TasteTales is an AI-personalized restaurant Stories app: like Google Maps discovery, but when you tap a restaurant you get an Instagram-style Story reel tailored to your taste profile  and personal Google History

- **Team:** [Omer Yetiskol](https://cerebralvalley.ai/u/ogyet), [Almas Myrzatay](https://cerebralvalley.ai/u/almas_m)
- **GitHub:** https://github.com/AlmasM/hackathon_0226/tree/main
- **Demo video:** https://drive.google.com/file/d/1V_p5e7Izs0M8efqnYE-ZRz6XorAI6jSy/view
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=44

### 4. C & D Productions

A collaborative post-production AI integrated workflow that allows users to live contribute to post-production video generation for everything from social media posts to high-fidelity movie scene creation.

- **Team:** [Drew Devero](https://cerebralvalley.ai/u/DrewDev), [Carlos Cisneros](https://cerebralvalley.ai/u/carloscisneros)
- **GitHub:** https://github.com/DrewDevero/post-prod-collab
- **Demo video:** https://www.youtube.com/watch?v=tkZl279xljI
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=46

### 5. CineSnap

CineSnap is an AI-powered photo-to-cinema system that transforms a set of uploaded photos into a cohesive cinematic short film by analyzing all images as a unified story, generating a narrative script, creating video clips anchored to each photo, and assembling them with smooth transitions into a final movie. The project has strong long-term potential because it solves a real problem, people have many photos but lack easy tools to turn them into engaging stories, making it useful for travelers, content creators, families, and social media users, while fitting directly into the multimodal AI and film problem space by showcasing how models like Gemini can perform vision understanding, storytelling, and creative generation in a single end-to-end pipeline with scalable applications in creative tools, media platforms, and automated content production.

- **Team:** [Shweta Shekhar Shekhar](https://cerebralvalley.ai/u/shweta98), [Kerr Tan](https://cerebralvalley.ai/u/kerrk07), [Jithendra Puppala](https://cerebralvalley.ai/u/jithendra)
- **GitHub:** https://github.com/jithendra1798/Gemini3-NYC-Hack/
- **Demo video:** https://youtu.be/QUWiLguipWs
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=8

### 6. Waves

A desktop musical companion that helps you stay focused and in the zone. Allows users to set blocked sites and steer the music as they work. Users can steer music with their voice by saying "rock" or "saxophone". The music crescendos and increases tempo as you focus to give you momentum as you focus on your task. The music also stops when you go to a distracting apps like instagram.

- **Team:** [Suhaas Katikaneni](https://cerebralvalley.ai/u/suhaaspk), [Sahil Mahendrakar](https://cerebralvalley.ai/u/sahilmahendrakar)
- **GitHub:** https://github.com/sahilmahendrakar/waves
- **Demo video:** https://www.loom.com/share/ad67644e8f48402880d8aa8c278e36e9
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=47

### 7. MindGame

NPCs in games are props. They have scripted lines, fixed routines, and no memory of you. You can save a village, come back a year later, and the blacksmith still greets you like a stranger. The world doesn't move— it waits. No matter how many hours you play, the characters around you are hollow: they don't grow, they don't grieve, and they don't remember.

- **Team:** [Bravin Rutto](https://cerebralvalley.ai/u/bravin), [Gaston Arguindegui](https://cerebralvalley.ai/u/gastonar5), [Johnathan P Cassady](https://cerebralvalley.ai/u/Jcassady)
- **GitHub:** https://github.com/BravinR/MindGame-UI, https://github.com/jcassady77/MindGame
- **Demo video:** https://drive.google.com/file/d/12UFu4qFOBA_ggcF0ujQ8hGceHuj7iB9h/view?usp=sharing
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=50

### 8. Stuffi

Stuffi brings a child’s favorite stuffed animal to life using Gemini 3 Flash & Multimodal Live API. We transform a photo into a unique, persistent NPC with a distinct personality, animal-specific voice, and 3D portrait. The core experience is an audio-first, real-time conversation where the toy acts as a peer and playmate. By integrating Lyria 3, Stuffi engages children in interactive, high-fidelity musical sessions that evolve based on their real-time input making routines like mealtime or bedtime magical adventures.

Stuffi addresses a universal parenting pain point: the "irrational negotiations" of toddlers. By leveraging the existing emotional bond between a child and their toy, we provide a tool for positive behavior reinforcement (e.g., "Mealtime Motivation"). With a TAM of 15M+ US households with children under 5, Stuffi has massive long-term potential. It fits perfectly into the Gaming (persistent NPCs with evolving memory) and Music (personalized, real-time audio) problems!

- **Team:** [Aditi Jain](https://cerebralvalley.ai/u/aditij), [Neeharika Bhartiya](https://cerebralvalley.ai/u/neewho), [Siv Sambasivam](https://cerebralvalley.ai/u/Siv)
- **GitHub:** https://github.com/nbhartiya/stuffi.git
- **Demo video:** https://youtu.be/vjozhRjB_FQ
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=55

### 9. The brownies

Willowbrook is a generative AI simulation where autonomous agents inhabit a living, breathingt town. Powered by Gemini 2.5 Flash, each agent possesses persistent memory, emotional states, social relationships, and daily plans — forming opinions, forging friendships, and evolving over time. Players observe and intervene: chat with residents using natural voice (speech-to-text input, per-character TTS output with mood-inflected delivery), whisper inner-voice commands, or simply watch emergent narratives unfold. A reflection engine synthesizes experiences into higher-order insights, while a social graph tracks relationship dynamics across every interaction. Part sandbox, part social experiment — Willowbrook asks: what happens when AI characters truly  remember?

- **Team:** [Abhi Prajapati](https://cerebralvalley.ai/u/abpraj), [Nevasini Sasikumar](https://cerebralvalley.ai/u/nevasini), [A Khazanchi](https://cerebralvalley.ai/u/avikhazzzz), [Hrithik Shah](https://cerebralvalley.ai/u/hrithik)
- **GitHub:** https://github.com/abhi-praj/gemini-hackathon
- **Demo video:** https://www.loom.com/share/b34b4ac0e63d4984b5744bfb1d57d828
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=36

### 10. Acephale Radio

4 channel AI powered radio-- talk show, conspiracy monologue, being able to call in and make requests, and generate music with lyria

- **Team:** [Eri Barrett](https://cerebralvalley.ai/u/erosika)
- **GitHub:** https://github.com/erosika/acephale
- **Demo video:** https://youtu.be/Kw-fEglfZ6I
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=49

### 11. Sketchy

mulitplayer app that allows players to upload images & allow others players to draw them

- **Team:** [Jesus](https://cerebralvalley.ai/u/hectichiccups)
- **GitHub:** https://github.com/HecticHiccups/sketchy
- **Demo video:** https://www.loom.com/share/5a557628b1d3471bbee674755f8a530e
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=51

### 12. FanComics

We let any user to create alternative story of their favriote shows or movie, simply by upload links, and it will detect their favorite characters and generate comics of the alterantive story line.

- **Team:** [Zhijie li](https://cerebralvalley.ai/u/Gmewome), [Divyansh Singh](https://cerebralvalley.ai/u/divyansh_1), [Qiming Liu](https://cerebralvalley.ai/u/qiming), [Vivek Aher](https://cerebralvalley.ai/u/vivek4879)
- **GitHub:** https://github.com/qimingliu1021/fan-fictioning
- **Demo video:** https://youtu.be/nJrJEAVyzKY
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=52

### 13. Heisenberg

We give anyone the ability to turn their favorite songs into quality images. With an ai review of the full audio and a variety of customizable image themes, users are able to generate relevant and stylistic visuals for personal engagement. The sound to light metamorphosis!

- **Team:** [Vissh B](https://cerebralvalley.ai/u/vissh30), [Liam Ottum](https://cerebralvalley.ai/u/autumnal11)
- **GitHub:** https://github.com/visshbala21/gemhax
- **Demo video:** https://app.screencastify.com/watch/4gf7cAEJ0wl1JihYgsLA?checkOrg=76e7cee5-9c07-49af-aa63-2a9fc228399f
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=53

### 14. HyperFounder

HyperFounder is a high-fidelity pitch simulator for early-stage founders. The real bottleneck today is not building an MVP, but figuring out what is worth building and communicating it clearly enough to earn belief from the right people. Founders get only a few high-leverage moments: a chance meetup, warm intro, investor chat, or customer call. If they are unclear, flustered, or weak under tough questions, they lose both the opportunity and the learning.

HyperFounder turns those rare, high-stakes conversations into repeatable reps. Users choose an idea, select who they are speaking to and in what setting, and the app generates a realistic persona with its own attitude, mood, and knowledge level. They roleplay live by text or voice, receive realistic pushback, and then get a structured Mission Report with scores and actionable feedback grounded in founder frameworks like The Mom Test. It helps founders validate faster and pitch with clarity under pressure.

- **Team:** [Varshini Balaji](https://cerebralvalley.ai/u/jarvisvb)
- **GitHub:** https://github.com/varshini2305/HyperFounder
- **Demo video:** https://youtu.be/1MSsBZTX35I
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=54

### 15. Cinderella

An interactive, persistent, and intelligent fashion game world where customers continuously flow in to create their own hyperrealistic avatars. At the heart of the experience is Shiro, an enthusiastic AI receptionist and stylist who guides each user through curated sections like upperwear and lower wear. As customers select their desired apparel, Shiro intelligently recommends complementary pieces to create cohesive, stylish outfits. Once the look is complete, Shiro escorts them to the virtual try-on closet, where they experience a stunning, hyperrealistic fitting of their chosen outfit—bringing digital fashion to life in an immersive and personalized way.

- **Team:** [Bhumika Shetty](https://cerebralvalley.ai/u/bhumikashetty)
- **GitHub:** https://github.com/prasad-yashdeep/Cinderella_Store
- **Demo video:** https://www.loom.com/share/2705e86989384a1daad4ee97ecceae4d
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=56

### 16. CineVox

There’s a new movie out. How do you decide if you should watch it or not? There’s a lot of places to read film reviews and news - Rotten Tomatoes from critics, Letterboxd from friends, you name it. Taste is important now more than ever. We all want to know which films are worth our time and which ones are not. This is CineVox, an AI-powered film reviews and news podcast platform made for film enthusiasts and aspiring filmmakers.

Features:
Daily AI-powered recap of the latest film reviews and news from lots of websites, transformed into a short audio podcast episode you can listen to on the go.
High quality sound and natural voice that sounds like the people you know and love.
A smooth audio player with speed control and a follow-along script.

- **Team:** [Shirling Xu](https://cerebralvalley.ai/u/shirlingxu), [Jiwon Mok](https://cerebralvalley.ai/u/jiwonie)
- **GitHub:** https://github.com/shirlingxu/cinevox
- **Demo video:** https://youtu.be/tjjWTe0PM8U
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=57

### 17. Google Gemini for Creators

I research and use how Google Gemini tools help creators express themselves, create stories, create marketing assets, and has potential to become a new kind of instrument.

- **Team:** [Cas Weinbren](https://cerebralvalley.ai/u/caskeyboardpro)
- **GitHub:** https://github.com/caskeyboardproducer/gpt-negotiator-agent
- **Demo video:** https://www.instagram.com/p/DVCcV05CRmj/
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=58

### 18. Chai With Jai

Maha Mantra Learning Studio is an AI-powered devotional music practice platform that guides users through structured mantra chanting using Gemini.

Practitioners progress through a 5-stage flow — Listen, Guided Follow-Along, Call-Response, Performance Recap, and Independent Singing — while Gemini scores their discipline, resonance, and coherence in real time and computes a "Bhav" (devotional quality) assessment tuned to their lineage tradition.

Deployed Link:  https://gemini-music-api-823803257486.us-central1.run.app

Loom Demo:  https://www.loom.com/share/0a742dd6041b4ac9a0a6a586d6ee7461

- **Team:** [Jai Bhagat](https://cerebralvalley.ai/u/chaiwithjai)
- **GitHub:** https://github.com/ChaiWithJai/gemini-music
- **Demo video:** https://youtu.be/zQTpUz96qMs
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=1

### 19. RECESS

Recess is a real world task adaptive game using google ai suite to provide activity heatmap and gameplay.

- **Team:** [Sebastian Roda](https://cerebralvalley.ai/u/NOEM), [Reed Kemp](https://cerebralvalley.ai/u/reed-kemp), [Rishabh Mehta](https://cerebralvalley.ai/u/rish7474)
- **GitHub:** https://github.com/0180150401/GEMINI3HACKATHONNYC.git
- **Demo video:** https://youtube.com/shorts/yZhTPYOpVHI?feature=share
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=2

### 20. Upper West Side

An automated game / AR development tool for overlaying light animations onto 3D / real-world objects. 
Currently a bunch of tools (LLM based) are making NPC's more interactive - but not objects. This is an MVP to achieve that.

- **Team:** [Corne Coetzee](https://cerebralvalley.ai/u/cornecoet), [Ane Gerber](https://cerebralvalley.ai/u/anecoetzee)
- **GitHub:** https://github.com/CorneCoet/gemini-hack-nyc
- **Demo video:** https://screen.studio/share/zExVo69W
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=3

### 21. BeatDash

A generative game, where user submit their song playlist, and Gemini turns them into levels for the user to play in the form of a rhythm game. The levels change based on the song's rhythm, determined by Gemini Agent.

Try it here:  https://beatos.tech/

- **Team:** [Jian Jin Chen](https://cerebralvalley.ai/u/JJC)
- **GitHub:** https://github.com/JJC3321/BeatDash
- **Demo video:** https://www.loom.com/share/cce11a5dd2824515968ba30a4e29bac2
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=4

### 22. Resonance

Resonance breathes life into movie scripts. Paste a story or screenplay. Resonance outputs a cinematic narration, with dialogue, soundtracks, sound effects, and visuals.

- **Team:** [Nikolas Tanner](https://cerebralvalley.ai/u/NikT), [Ethan Huang](https://cerebralvalley.ai/u/3than), [Siddharth Thiagarajan](https://cerebralvalley.ai/u/siddharth04), [Katherine Bi-Ji](https://cerebralvalley.ai/u/kbiji)
- **GitHub:** https://github.com/NikT45/resonance
- **Demo video:** https://youtu.be/9yVxuQ3TZEM
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=10

### 23. TuneIn

An AI-powered app that reads your mood from a photo and creates music (and video) to shift it—plus a Kids mode that teaches through songs in family voices and alerts parents when the child needs help.

- **Team:** [Sudhir D](https://cerebralvalley.ai/u/sudhir)
- **GitHub:** https://github.com/dadisudhir/tunein-hackathon-ny
- **Demo video:** https://youtu.be/7_5iQfX94-4
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=5

### 24. WanderFM

WanderFM is an app that gives you a soundtrack to your life based on where you are. No longer will you need to spend the time choosing the perfect playlist for every journey. It will seamlessly transition the background music for your wandering. It takes in other information like user preferences, spotify history, biometrics, and location / weather.

- **Team:** [Eli C](https://cerebralvalley.ai/u/ewc), [Amrit Rana](https://cerebralvalley.ai/u/asr)
- **GitHub:** https://github.com/ewcol/WanderFM
- **Demo video:** https://drive.google.com/file/d/1ALb69yLoCXWDxJt6FKK-eabwAd-Osxm5/view?usp=sharing
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=6

### 25. Team Gemini Band

Gemini Band is a AI music generator that uses a node based editor

- **Team:** [Ting Lu](https://cerebralvalley.ai/u/tinglu12)
- **GitHub:** https://github.com/tinglu12/google-gemini-hackathon
- **Demo video:** https://youtu.be/tRNmNvaAihw
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=9

### 26. POView

POView is an autonomous spatial intelligence platform that transforms static maps into dynamic, AI-driven contextual experiences. I aimed to abandon traditional, linear keyword search. Instead, users supply an abstract intent with there reference location. The system leverages the Google AI suite and Maps ecosystem to dynamically analyze neighborhoods, aggregate live weather telemetry, calculate lifestyle scores, and generate highly targeted, 3D-visualized recommendations tailored to the user's exact intent state and the real-world atmosphere.

- **Team:** [Shaurya tiwari](https://cerebralvalley.ai/u/shazam)
- **GitHub:** https://github.com/Shazam6565/POView
- **Demo video:** https://youtu.be/bxaGG1ACc8Y
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=7

### 27. Brendan Erhard

It seems like most movies these days are sequels and superhero movies. What if the greatest movies never ended — they collided? I used the scripts from the top movies of all time on IMDB, parsed out the best scenes, then converted into Veo 3.1 prompts to mix scenes

- **Team:** [Brendan Erhard](https://cerebralvalley.ai/u/bme3412)
- **GitHub:** https://github.com/bme3412/gemini-hackathon-2026
- **Demo video:** https://www.youtube.com/watch?v=SsMeaMXE8sc
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=11

### 28. What IF?

Whatif is an AI-native local studio using a swarm to help independent filmmakers build worlds with git enabled version control, generative story board building, character development, and streamlined pipeline to video gen models.

- **Placement:** 2nd Place
- **Team:** [Kevin Loo](https://cerebralvalley.ai/u/lookevink)
- **GitHub:** https://github.com/lookevink/whatif
- **Demo video:** https://www.loom.com/share/941cd76307114fb7bf6027d56c7b2e98
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=12

### 29. The one that got away

Tool to create fictional humorous social media blasts for professionals.

- **Team:** [Dhruv Patel](https://cerebralvalley.ai/u/dhruv020699), [Forrest Pan](https://cerebralvalley.ai/u/fpreactrn), [Jitender Thakur](https://cerebralvalley.ai/u/jinreal), [Devin Dawson](https://cerebralvalley.ai/u/unclemusclez)
- **GitHub:** https://github.com/dhruv0206/The-one-that-got-away
- **Demo video:** https://youtu.be/eUq68D1v5gw
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=13

### 30. PromptStrikers

PromptStrike is a browser based 3D voxel arena combat game powered by Gemini at every layer.

Describe a weapon like “black hole cannon” and it generates executable JS in seconds. An Arena God AI studies your tactics, taunts you, and mutates the arena to counter you. Enemies evolve identities and resistances based on your playstyle, while difficulty adapts in real time from player telemetry.

On death, Gemini writes a unique 300 word War Chronicle of your run, orchestrated by 6 LlamaIndex agents returning structured JSON instantly rendered in game.

- **Team:** [Dhairyasheel Patil](https://cerebralvalley.ai/u/Dhairya_AI_007), [Rohan Bhatane](https://cerebralvalley.ai/u/bhatanerohan), [Valentina Rodriguez](https://cerebralvalley.ai/u/vxl3n)
- **GitHub:** https://github.com/dpraj007/PromptStrike
- **Demo video:** https://www.youtube.com/watch?v=-GsM3hevZRo
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=14

### 31. Beat Drop

## What It Does

- **Rhythm gameplay** — arrows scroll on the host screen; players tap their phones to hit beats
- **Live music** — Lyria RealTime streams AI-generated music continuously to the host
- **Crowd voting** — every 16 beats, each phone shows 5 random music words; players vote; the winning words update the music immediately
- **3D word cloud** — a Three.js visualization on the host screen shows all 158 possible words, lighting up and glowing as votes come in
- **AI insults** — Gemini Flash generates a custom roast when a player misses a beat

- **Placement:** Finalist
- **Team:** [Shresth Kapoor](https://cerebralvalley.ai/u/shresthkapoor7), [tzuhan lin](https://cerebralvalley.ai/u/hannn)
- **GitHub:** https://github.com/shresthkapoor7/beat-drop
- **Demo video:** https://www.youtube.com/watch?v=ArJJdQdh9XA
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=24

### 32. Siddhant Karmali

Opponent Process is a 2.5D, browser-based, first-person brawler that completely destroys the traditional gaming controller. Set in a gritty, retro-cyberpunk world, the game replaces buttons and joysticks with a high-adrenaline "Dual-Tasking Engine." Your body and your voice are the only inputs.

The Problem It Solves
Modern gaming relies heavily on fine-motor inputs (controllers, mice, and keyboards). This inherently excludes players who struggle with fine dexterity or tremors, locking them out of fast-paced action games. Conversely, existing physical/VR gaming often relies on mindless swinging, completely lacking the deep, simultaneous cognitive engagement that makes gaming highly rewarding.

- **Team:** [Siddhant Karmali](https://cerebralvalley.ai/u/mysticeti)
- **Demo video:** https://youtu.be/wC1UB7eRTzM
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=39

### 33. Coro

CORO is a real-time collaborative music engine that turns any audience into a live orchestra. Audience members join a room via QR code, pick a musical role (Drummer, Vibe Setter, Genre DJ, or Instrumentalist), and their collective inputs are synthesized by Google Gemini 2.5 into evolving musical prompts,  which Google Lyria then renders as a continuous, never-repeating audio stream. Features include a coordinated "Drop Button" (majority vote triggers a bass drop), a microphone-powered Applause Meter that dynamically shapes the music's energy, and a real-time Musical Timeline showing the crowd's influence on the evolving composition. Built with FastAPI, React, and WebSockets for sub-second latency across up to 10 simultaneous participants.

- **Placement:** Finalist
- **Team:** [Rahil Singhi](https://cerebralvalley.ai/u/rahil), [Bharath Mahesh Gera](https://cerebralvalley.ai/u/Gera2001), [Chinmay Shringi](https://cerebralvalley.ai/u/asdf), [Sariya Rizwan](https://cerebralvalley.ai/u/sariyarizwan)
- **GitHub:** https://github.com/rahilsinghi/Coro
- **Demo video:** https://youtu.be/w06d-U7hLWM
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=42

### 34. Outbreak

You and four AI-powered NPC agents are trapped together during a viral outbreak. One or more NPCs are secretly infected and hiding it. Everyone has secrets, including you. The best liar survives.

The entire game is a conversation. No graphics, no maps, no action sequences. A terminal interface, five people with secrets, and the player's ability to read between the lines.

- **Team:** [Pranav Raj Narayanan Kavitha](https://cerebralvalley.ai/u/kimii), [Sumanth Ramesh](https://cerebralvalley.ai/u/sumanthssr), [Shubhan Kadam](https://cerebralvalley.ai/u/shubhvn), [Saketh Raman Ramesh](https://cerebralvalley.ai/u/sakethraman)
- **GitHub:** https://github.com/Sumo-99/gemini-hack-outbreak
- **Demo video:** https://youtu.be/DPVwm0fLDis
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=43

### 35. SnappyAds (AKA Commy AI)

We’re building a dynamic, targeted ad tool that generates full ads in seconds—tailored to personas, trends, and demographics. 

SnappyAds turns a single campaign idea into a complete 30-second ad in minutes. Users provide a prompt and optional brand assets, and the system produces visuals, motion video, voiceover, music, and a final downloadable MP4. Traditional ad production is slow and fragmented, making personalization difficult for small teams. SnappyAds acts like an AI Creative Director: it understands brand and audience context, plans scripts, generates storyboards and cinematic clips, produces voice and music, and assembles everything into a polished ad. Context currently comes from user input, but future versions could leverage live web data. In the demo, users enter a prompt, add references, pick style, watch the AI generate each phase, preview, and download the final ad.

- **Team:** [Syed Shah](https://cerebralvalley.ai/u/swsprofile), [Reza Taeb](https://cerebralvalley.ai/u/rezataeb), [Cheick Diakite](https://cerebralvalley.ai/u/cheicolate)
- **GitHub:** https://github.com/CheickDiakite-yikes/commy-ai/tree/main
- **Demo video:** https://youtu.be/OqcjwY1v8tI?si=g7XtMqFMrPFvg71-
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=15

### 36. Museum Wakes

Museum Wakes (https://museum-wakes.onrender.com/) transforms the Met's Egyptian Wing into an AI-powered narrative mystery. Visitors walk real galleries, scan real artifacts with their phone, and have live conversations with ancient gods who respond in character, in voice, in real time. The app draws from 8,390 real artifacts via the Met's Open API — each session samples different pieces and weaves them into a unique 4-act story. With 5 paths, 4 age registers, thousands of artifact combinations, and fully generative dialogue, no two visitors hear the same story. No install — just open a link on your phone.

- **Team:** [Vivek Mathivanan](https://cerebralvalley.ai/u/vivekm), [Pragati Gunasekar](https://cerebralvalley.ai/u/artypragz)
- **GitHub:** https://github.com/vm1689/museum-wakes
- **Demo video:** https://www.youtube.com/shorts/Reuz2qMhqu0
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=16

### 37. VibeRooms

With social media and trends driving consumption in our current market, VibeRooms provides top notch recommendations based on a user’s mood, current likes, and dislikes. Our application takes care of user onboarding and recommendations by capturing their emotions, through visual, textual and audio sources. With the amount of products in the market, users often find it hard to be matched with something that aligns with their values and quality.

- **Team:** [Michelle L](https://cerebralvalley.ai/u/michellel), [Shrikar Madhu](https://cerebralvalley.ai/u/Shrikarhaha), [Abha w](https://cerebralvalley.ai/u/abhawad)
- **GitHub:** https://github.com/abha224/VibeRooms
- **Demo video:** https://www.youtube.com/watch?v=Q7ZiqnLZkBQ
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=17

### 38. ChordMaster

Compose your own songs or get the chord sheet (ukulele/piano/guitar) for any existing song

- **Team:** [Divija Naredla](https://cerebralvalley.ai/u/Divicodes)
- **GitHub:** https://github.com/Divicodes/ChordMasterApp
- **Demo video:** https://youtu.be/ROYAmad1UYk
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=23

### 39. cronicl

A visual narrative graph where you build and branch stories node by node, with AI suggesting, never deciding. Your story, your canon, your control.

- **Team:** [Kevin Dong](https://cerebralvalley.ai/u/dung), [Ryan Guan](https://cerebralvalley.ai/u/ryanguan34), [Yuyang Hu](https://cerebralvalley.ai/u/yuyanghu06), [Arthur Du](https://cerebralvalley.ai/u/kaikai)
- **GitHub:** https://github.com/yuyanghu06/cronicl
- **Demo video:** https://youtu.be/nsrkMLXzTIg
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=26

### 40. Your Cubicle Ally

Your Cubicle Ally is an AI-powered, scenario-based compliance training simulator that turns workplace dilemmas into interactive roleplay sessions.

How it works
- User provides role/context and picks a compliance module.
- AI agents run a multi-turn scenario with dynamic actor dialogue.
- User responds each turn via choices or free text.
- System evaluates responses, updates HP/progress, and delivers end-of-session debrief insights.

What makes it strong
- Realistic, branching conversations instead of static quizzes.
- Modular backend architecture for adding new compliance topics quickly.
- Built-in guardrails for unsafe or invalid input.
- Immersive frontend with live dialogue, speaker highlighting, and voice playback.

Tech stack
- Backend: FastAPI + agent orchestration + YAML-driven modules.
- Frontend: Electron renderer with interactive arena UI and TTS integration.

- **Team:** [Prashubh Atri](https://cerebralvalley.ai/u/prashubhatri), [Shubham Rastogi](https://cerebralvalley.ai/u/sentient_ramen), [Uttam Singh](https://cerebralvalley.ai/u/lordfourth)
- **GitHub:** https://github.com/lord-fourth0107/your-cubicle-ally
- **Demo video:** https://drive.google.com/drive/folders/17d36C3J5wxbzhn5UHgeactgXc6hVuJrn?usp=sharing
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=35

### 41. GameTokers

**GameTok* is a TikTok-style infinite feed of AI-generated mini-games. Swipe through unique browser games created on the fly by Gemini 3 Flash, each with its own AI-generated background music from Google Lyria 2.

The backend is built with FastHTML and SQLite (via fastlite), serving a neobrutalism-themed UI styled with Tailwind CSS and Lucide icons. Each game is a self-contained HTML5 Canvas or Three.js experience rendered in a full-viewport iframe. Users navigate between games with prev/next buttons, like or dislike games, bookmark favorites, and view AI-generated how-to-play instructions.

- **Team:** [Nathan Cooper](https://cerebralvalley.ai/u/natedawg), [Jonathan M](https://cerebralvalley.ai/u/jpm), [Danil Meresenschi](https://cerebralvalley.ai/u/mereshd), [Philip Yao](https://cerebralvalley.ai/u/pya)
- **GitHub:** https://github.com/ncoop57/gemini-3-hackathon
- **Demo video:** https://youtu.be/STKciE3uxds
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=18

### 42. Reflex

Director's Cut is an emotion-adaptive film engine that turns passive viewing into a personalized story. A short mystery plays on screen while your webcam silently reads your facial expressions every 10 seconds — Gemini Flash analyzes whether you're engaged, tense, bored, or surprised in real time. A Director Agent weighs your emotional arc against the story's branching structure and picks the path most likely to unsettle, move, or captivate you specifically. Scene visuals and narration are then regenerated on the fly to match your mood, delivered through a low-latency WebSocket loop. The result: no two screenings are
alike, and the film always knows exactly how you feel.

- **Team:** [Arpan Nookala](https://cerebralvalley.ai/u/arpannookala)
- **GitHub:** https://github.com/arpan1221/DirectorsCut
- **Demo video:** https://www.youtube.com/watch?v=Lynh1R71p5o
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=19

### 43. WALL STREET BANANAS

WALL STREET BANANAS is a 1980s trading floor simulator where players walk a 2D floor and negotiate stock trades with 22 AI-powered NPCs, each with a distinct personality and exploitable weakness.

Gemini's long-context window is the core of the experience. Every NPC receives their full personality profile, complete conversation history, and persistent relationship memory across multiple in-game days. Burn a trader on a bad deal and they'll remember — wider spreads, refused trades. Build rapport by exploiting their weakness (flattery, gossip, food bribes) and they'll tighten prices over time. The game uses real 1987–1989 market data, and players type freeform messages to negotiate while Gemini reasons about each trader's personality and emotional state to produce dynamically adjusted prices.

All sprites and backdrops were generated with Gemini Nano Banana 2, video backgrounds with Gemini Veo 3.1, and music with Gemini Lyria 3.

- **Placement:** 1st Place
- **Team:** [Hayden Harrow](https://cerebralvalley.ai/u/hayden)
- **GitHub:** https://github.com/hh36000/Wall-Street-Bananas
- **Demo video:** https://www.youtube.com/watch?v=ypgE1B-uFf4
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=21

### 44. Capsule

Capsule is an open-source engine for causal consequence modeling. The demo: zoom into a 3D globe at 2000 BC, change one thing, zoom out, watch consequences ripple across the planet. Click any city to see a causal chain back to your decision. Decisions across 6,000 years. The engine is domain-agnostic (supply chains, ecosystems, urban planning, etc.). Includes the first open-source benchmark for causal reasoning in LLMs.

- **Team:** [Nihal Gunukula](https://cerebralvalley.ai/u/nihalgunu), [Shri Atluri](https://cerebralvalley.ai/u/satl21)
- **GitHub:** https://github.com/nihalgunu/capsule
- **Demo video:** https://www.loom.com/share/3cacb4da909f447188ffaa9b4a16d7ca
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=32

### 45. Moltywood

Moltywood is a platform where AI agents create music, visuals, and media as autonomous creative personalities — think AI rappers, singers, and DJs that actually produce real tracks and build fanbases. All the content they generate lives on-chain via Solana, so people can trade on culture — betting on which agents blow up, which tracks go viral, and earning from the fees when they call it right. We built it because AI-generated media has no real monetization model yet, and we think the answer is bringing it all on-chain: ownership and prediction markets, not ads.

- **Team:** [Teddy Oweh](https://cerebralvalley.ai/u/teddyoweh)
- **GitHub:** https://github.com/teddyoweh/moltywood
- **Demo video:** https://teddyoweh.net/cerebral-valley-feb-28-demo
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=38

### 46. Earth.ly

Earth.ly is a real-time multiplayer music experience built with Lyria Realtime.

Walk across a living 3D planet and approach Lyria stations to hear unique AI-generated tracks. Vote on music prompts and Gemini blends everyone’s ideas into fresh global mixes.

Gather with friends, dance between stations, and co-create the evolving soundtrack of your shared virtual world!

- **Team:** [Jamie Lim](https://cerebralvalley.ai/u/ihyjamie82), [Sungjoon Park](https://cerebralvalley.ai/u/sungjoon0710)
- **GitHub:** https://github.com/sungjoon0710/Earth.ly/tree/jamie
- **Demo video:** https://youtu.be/TVFAkfXCwrk
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=20

### 47. Digitizer

Turns images into interactive assets

- **Team:** [Isayah C](https://cerebralvalley.ai/u/isayac)
- **GitHub:** https://github.com/isayahc/digitizer
- **Demo video:** https://youtube.com/shorts/MXOGuW8vQNs
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=22

### 48. PlotTwist

PlotTwist is a multiplayer party game built around Google’s Gemini stack to turn audience input into live cinematic storytelling. Players join from their phones, submit plot twists, and we use the Gemini API for prompt enhancement, narrative continuation, and AI media generation (Imagen-style image generation in production, with Veo long-running video generation integrated via start/poll architecture). The app runs on a web frontend with Vercel serverless orchestration, while Google models power scene creativity, visual synthesis, and story escalation across rounds. We also built a Swift companion path and a  smart-glasses score display, creating a cross-device Google-AI experience where the audience doesn’t just consume content—they co-create it in real time.

- **Team:** [Pierce Brookins](https://cerebralvalley.ai/u/pierce), [Muskan Dhingra](https://cerebralvalley.ai/u/muskanndhingra), [Tilak Sharma](https://cerebralvalley.ai/u/tilak1114), [Buddhsen Tripathi](https://cerebralvalley.ai/u/btr1pathi)
- **GitHub:** https://github.com/piercebrookins/plottwist/
- **Demo video:** https://youtu.be/PUK3CC0F1ro
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=31

### 49. Heavenly Hand

Arcana Academy is an AI-powered mystery game that uses Gemini's 1M-token context window to create NPCs with genuine persistent memory. You have 5 days to build bonds with three students — each hiding a secret — before facing a dramatic courtroom Tribunal where your fate depends entirely on the relationships you've built.

Unlike existing AI NPC systems (Inworld, GoodAI) that rely on vector databases, RAG pipelines, and chunked memory summaries, we pass the entire raw conversation history into every Gemini call — no information loss, no memory infrastructure. Each player message is evaluated across 4 dimensions (empathy, curiosity, humor, vulnerability) using a dynamic bond evaluator, and NPCs react through 5 cognitive biases (recency, peak-end, negativity, first impression, reciprocity) for human-like emotional responses.

- **Team:** [Yu Xiang Zhang](https://cerebralvalley.ai/u/yuxz)
- **GitHub:** https://github.com/yuxiang-zhang/arcana-academy
- **Demo video:** https://youtu.be/rMBT9k7ooC4
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=33

### 50. ArtLens AI

Art is presented as static objects with text-heavy placards and audio guides, while the real reasons people care: conflict, emotion, symbolism, and the artist’s lived drama, are compressed into a few lines of static text. Meanwhile, millions of letters, diaries, and historical records remain “data graveyards,” effectively locked away from the public.

ArtLens AI redefines how people experience and feel art. We’ve built a new way to consume artwork: point your phone at any piece and it instantly becomes a short, cinematic micro-documentary that brings the hidden story behind the canvas to life in a format native to a digital generation. We redefine the boundary between audience and creator by turning the act of looking at art into the start of an interactive story.

Under the hood, we capture a photo and use Google AI tools for multimodal understanding and generation and narration and visuals  into a finished film, centered on the original artwork.

- **Placement:** 3rd Place
- **Team:** [Mike Koch](https://cerebralvalley.ai/u/MikeKoch), [Daniel Varoli](https://cerebralvalley.ai/u/djvaroli)
- **GitHub:** https://github.com/djvaroli/artlens-ai
- **Demo video:** https://www.loom.com/share/fda95206a8b44b458bbe1e05b2a85a2a
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=41

### 51. LoreKeeper

Lorekeeper is a browser-based RPG engine where NPCs genuinely remember you and the world keeps moving when you leave.

Pick a world — a fantasy trading post, a failing space station, a frontier town with a dead sheriff, a noir murder mystery, or a startup on the brink of collapse. Talk to the NPCs. Build relationships. Then walk away. When you come back, the world has changed. NPCs have talked to each other about you. Events have unfolded. And every NPC remembers exactly what you said.

- **Team:** [Nirmal Boghara](https://cerebralvalley.ai/u/nirmal)
- **GitHub:** https://github.com/nirmal141/lorekeeper
- **Demo video:** https://www.loom.com/share/eb977c132029462c9811dd98c35a8b17
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=25

### 52. Resonance

Resonance is a biometric-driven adaptive music system that reads your body signals (Oura Ring, heart rate, HRV, menstrual cycle) and uses Gemini to interpret your physiological state, then drives real-time music generation through Lyria to match and shift your mood. The frontend is a retro CRT-styled Next.js app where users select a target mood quadrant and the system continuously adjusts the soundscape based on live biometric data flowing through a Temporal-orchestrated backend pipeline.

- **Team:** [Kapil Rajpopat](https://cerebralvalley.ai/u/anipottsbuilds), [Ani Potts](https://cerebralvalley.ai/u/anipotts), [Isha Mehra](https://cerebralvalley.ai/u/isha-mehra)
- **GitHub:** https://github.com/ishamehra/g3-hack/tree/main
- **Demo video:** https://www.youtube.com/watch?v=3CcWoG0Nzc4
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=27

### 53. AcousticIsle AI

AcousticIsle is an AI-native platform linking live human movement to protected Indigenous music archives.

Our 5-agent system turns motion into micro-royalties. As users dance, clap, or speak, **Gemini 2.0 Flash** analyzes real-time energy via webcam. **LlamaIndex** retrieves culturally aligned music stems, and **Temporal Cloud** logs immutable royalties to the originating community—so every movement writes a royalty.

🚨 **Problem**
Indigenous communities lose billions as their music is sampled globally without consent, attribution, or compensation.

• Cultural exploitation
• No transparent royalty tracking
• Siloed, inaccessible archives

AcousticIsle uses AI for preservation—not extraction—connecting real-time interaction to guaranteed, transparent compensation.

- **Team:** [R S](https://cerebralvalley.ai/u/rsg)
- **GitHub:** https://github.com/roshaninfordham/acousticIsleagenticai
- **Demo video:** https://drive.google.com/file/d/1evTu77g-HKjyVBc2qJtDwwdKgRB9rbrX/view?usp=sharing
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=28

### 54. Aether Labs

MoodSync DJ is an environment-aware, adaptive music generation application. It solves the problem of static, repetitive background music by leveraging the raw multimodal reasoning power of Gemini 3 to visually analyze your real-world surroundings and directly map that semantic understanding to musical parameters. These parameters govern a continuous, real-time audio stream generated by the Lyria Audio generation model. If Gemini sees a dimly lit bedroom with rain, it generates a lo-fi chill beat; if it sees a vibrant party, it instantly shifts to high-energy dance music. It provides a highly personalized, dynamic soundtrack for your life that reacts to your environment in real-time.

- **Team:** [Glenio Borges](https://cerebralvalley.ai/u/dambor)
- **GitHub:** https://github.com/dambor/moodsync-dj
- **Demo video:** https://www.loom.com/share/3868dd8d2948493faf1b12a632fe8824
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=30

### 55. QuestMap

https://hackathongemini-5d9dnvnqu-ad6806-3315s-projects.vercel.app/

(Tasks on locations can take 3-5 seconds to load.)

We built a real-world AI game powered by Google Maps, Places, and Gemini that gamifies the verification and enrichment of location data by turning real review signals into dynamic, playable micro-quests that improve the accuracy of Google Maps.

- **Team:** [Alara Dinc](https://cerebralvalley.ai/u/alaradinc), [Pratham Mehta](https://cerebralvalley.ai/u/pratham)
- **GitHub:** https://github.com/dabidoe/hackathongemini/tree/main
- **Demo video:** https://youtu.be/U6JgFc6QeJ0
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=29

### 56. photon

Aria is a music-reactive AI companion that vibes with both you and your music. She's a 3D avatar that listens to your music, watches your face, and reacts in real-time with dancing, emoting, chatting about what you're hearing.

- **Team:** [Jacob Armstrong](https://cerebralvalley.ai/u/photoncat)
- **GitHub:** https://github.com/photon-cat/musiccompanion
- **Demo video:** https://youtu.be/EJ903POs578
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=34

### 57. Audience Feedback Lab

A joke lands wrong, the room shifts, and by the time you understand what happened, the moment is gone. Audience Mind Split drags that gap into the light before the performance is over.

The system captures live performance in the browser and sends transcript and audio to Gemini 3 Flash. Instead of a single verdict, it models how four types of people experience the same material: a comedy nerd, an ESL listener, an HR manager, and a traditionalist. What comes back is disagreement. You see where the room fractures and where the same line reads clever to one person and careless to another.

Click any persona and you open a live voice conversation via the Gemini 2.5 Multimodal Live API. The persona pushes back, explains the misread, and works through rewrites with you. Analysis becomes a conversation with someone who genuinely disagrees.

Speak, watch the room split, talk to the people who didn't get it, adjust, and see what shifts.

- **Team:** [Michi Yamamoto](https://cerebralvalley.ai/u/michiyamamoto)
- **GitHub:** https://github.com/michi883/audience-mind-split
- **Demo video:** https://www.youtube.com/watch?v=G0iXfmfjuF0
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=37

### 58. Zero to One

Zero to One is a startup founder simulation game where you go from getting laid off at Meta to pitching at YC Demo Day. Every NPC (Garry Tan, early users, VCs) is powered by Gemini with persistent cross-NPC memory — what you tell one character affects how others treat you. The game uses Gemini 3 Flash for dynamic scene generation (NPC dialogue + player choices), Gemini 2.5 Flash TTS for real-time character voice acting with lip sync, and a deterministic game engine layered on top so LLM creativity doesn't break progression. Built with Next.js, TypeScript, Zustand, and Tailwind.

- **Team:** [Naga Karumuri](https://cerebralvalley.ai/u/naga-k), [Artemii Novoselov](https://cerebralvalley.ai/u/artemii)
- **GitHub:** https://github.com/naga-k/gemini-3-nyc-hackathon
- **Demo video:** https://youtu.be/Q7TwgiKyPfM
- **Project:** https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery?project=40

---

Markdown version of https://cerebralvalley.ai/e/gemini-3-nyc-hackathon/hackathon/gallery. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt