Momentum

Momentum – Autonomous Multi-Agent AI Interview & GitHub-Based Technical Assessment Platform Momentum is an agentic AI platform that transforms technical interviews into interactive engineering assessments. It combines autonomous multi-agent orchestration, role-aware GitHub repository analysis, and practical coding evaluation. GitHub repository analysis and role-based bug injection Momentum connects to candidates' GitHub repositories, including private repositories through authorized access. It analyzes the codebase, understands project architecture, explores functions, and asks candidates about their specific roles and contributions. Based on their responses, the AI identifies code sections relevant to their claimed responsibilities and autonomously invokes its GitHub Code Bug Injector tool. The repository is cloned into an isolated container, creating a separate working copy. The tool introduces controlled bugs into selected code segments associated with the candidate's stated role. Candidates must investigate, identify, understand, and fix these bugs inside the isolated environment. The original repository remains untouched. This helps identify discrepancies between claimed contributions and demonstrated technical understanding while ensuring candidates are evaluated within their stated responsibilities. AI-triggered practical coding assessment Momentum features an interactive coding pad that AI agents can dynamically invoke as a tool during interviews. It evaluates programming syntax, SQL queries, algorithms, implementation skills, and logical reasoning through practical coding challenges. Multi-agent interview intelligence Multiple specialized AI interviewer agents collaborate through shared context and distinct roles. They conduct real-time, interruptible voice conversations, generate dynamic follow-up questions, adjust difficulty, challenge vague explanations, and explore contradictions. Evidence-based evaluation Momentum combines interview responses, practical coding performance, debugging results, repository evidence, and claimed responsibilities to generate a comprehensive technical evaluation. Security and isolation Repository assessments operate on isolated working copies, keeping the original repository separate from experimental modifications. Authorized private repository access enables assessment of real-world projects. Momentum's vision is simple: Don't just ask candidates what they have built. Let them demonstrate what they actually understand, own, and can engineer. Built for the Meta Global AI Developer Hackathon 2026 – Indie Developer Track.