Prakhar Goel@prakhargoel4554

AMD-Gemma Video Captioner is a containerized, production-ready video intelligence agent designed for the AMD Developer Hackathon. The project processes input video tasks to generate structured, creative captions in four distinct styles: Formal, Sarcastic, Humorous-Tech, and Humorous-Non-Tech. To optimize token consumption and performance, the pipeline implements: 1. Uniform Keyframe Sampling: Using OpenCV to extract exactly 10 frames across the video timeline, resizing them to 512px to cut visual token usage by 60%. 2. Temporal Grounding: Leveraging Kimi 2.6 VLM on the Fireworks AMD GPU cluster to map out an accurate chronological timeline of the video, preventing sequence hallucinations. 3. Single-Call Stylization: Querying DeepSeek V4 Pro using Pydantic JSON schema constraints to generate all 4 style captions in a single API call, reducing requests by 75%. 4. Dual-Key Schema Compatibility: Outputting both dash and underscore keys (e.g., humorous-tech and humorous_tech) to ensure the automated grading harness runs crash-free. The application is fully containerized using Docker, allowing stable headless execution in the evaluation environment.

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.