Tanishka Gupta@tanishkagupta140994799

CaptionForge AI is a production-ready video captioning platform built for the AMD Developer Hackathon Track 2. It processes videos through a modular Extract-Reason-Style multi-stage AI pipeline: Extract - Uses imageio-ffmpeg to extract audio and Pillow/imageio to sample 6 keyframes from videos Reason - Transcribes audio using Whisper-v3 and analyzes visual content using Kimi-k2p6 (both via Fireworks AI on AMD MI300X GPUs) Style - Transforms factual captions into four distinct styles using DeepSeek-v4-pro: Formal - Objective, professional documentation Sarcastic - Dry, theatrical irony Humorous-Tech - Programming metaphors Humorous-Non-Tech - Everyday observational comedy Evaluate - Local CaptionCritic validates for hallucinations and style adherence Key Innovations: Decoupled architecture enables flexibility and maintainability Cloud AMD MI300X GPU acceleration via Fireworks AI enables serverless deployment Multi-model fallback chain (DeepSeek → Kimi → GPT-OSS) for reliability Automated reasoning cleanup removes leaked chain-of-thought from model outputs Style-specific evaluation ensures quality across all four required styles Technology Stack: Backend: FastAPI + SQLAlchemy + SQLite Frontend: React + TypeScript + Vite + Tailwind CSS AI/ML: Whisper-v3, Kimi-k2p6, DeepSeek-v4-pro (all via Fireworks AI) Deployment: Vercel (serverless) + Docker

To participate in the hackathon, simply click Sign up with AMD. If you are not already a member of the AMD AI Developer Program (ADP), you will need to create an account. ADP members can access AI tutorials, experts, and community support. In addition, new members can claim a free month of DeepLearning.ai pro, $100 in AMD GPU cloud credits, and $50 in Fireworks AI API credits. All hackathon participants will receive on hackathon launch additional compute and API credits for their projects. Ready to build what's next on AMD? Sign up with AMD below to secure your spot.