Siddhant Choudhary
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

The project Aegis-home which i made was something i never expected to make when the building date started i had multiple ideas and all but i saw a smart home camera in my house which we never really used for an year i used to see it but i never got to think about its use case until on 29th Nov i got an idea to use this in my project to really make it worth it. Now i had the idea for a fun thing where my home camera suppose if i touch a object like my plushie which if i touch it could turn off/on my smart lights or anything which can be controlled within an network bascially first i got the idea to making steam deck but with physical objects but ofc it would require a Camera in order to maintain all that i was able to achieve that where it was able to tell the difference between touching and hovering over a object , what action does it do when we touch it (I added that to so that it can be configured to do anything suppose be a play pause button for my spotify ) and more. As a developer ofc this would be fun to use in daily life and then connect more devices which can be controlled ,Now i thought that its usecase is not much until i got the idea where people with disabilities could use this type of help. Since to see something we needs eyes why not make one i made my configured in such a way based on the disability or any physical health realted factor where the Camera if we setup smart home cameras and integrate our whole program it could help blind people to find its path using multiple cameras in house which could speak and help him assist forward or if the blind person wants to make you call someone from their contacts it can directly do that , Second where people who cant speak or listen i had the idea of where in front of the camera it understands sign langugage since many times when you are in home you may not be able to listen if the person is calling a member of the family or not and this could help to even communicate with their family by makign sign lanugage and tell our model to call my mom or something. Basically a full architecture using sam, spark and transcribe which could be integrated in smart cameras or even normal phone or laptop cameras which can act as all in one thing where we can talk to it , it can help us , it can make things futuristic where you can control things which obejcts and also coordinating with othe cameras ( i didnt have the time but i had the use case in mind to add security where the AI could use the cameras to check for night security in homes call 911 or 100 and sending reports and all if it detects and medical issue if on camera or on basis of voices ). This projects fallback can be that it takes personal data but for now since we had to use meta models so we cant do anything but if it were a real Product would have to train a LLM inside it and optimsed in such a way that it could handle multiple features. This was my idea to turn my camera which was there for no use to make it my buddy and helper.