Kiu Chun Woon@vicky_kiu

BudgetBrain Track 1 Champion Agent is a compact, Docker-ready AI reasoning system built for the AMD Developer Hackathon Act II Track 1 evaluation contract. The system reads tasks from /input/tasks.json and writes compliant answers to /output/results.json in the required [{"task_id","answer"}] format. It is designed to infer the hidden task category from each prompt, then choose the lowest-cost reliable solving path. The project combines deterministic local solvers with selective Fireworks model calls. Local modules handle math, sentiment, named entity extraction, logic, and code-debugging style prompts where rule-based reasoning can answer quickly and with zero tokens. For harder or ambiguous prompts, the router falls back to allowed Fireworks models using strict model gating, short timeouts, normalized outputs, and retry behavior that respects the hackathon constraints. The goal is strong leaderboard performance under unknown hidden tests: high accuracy, low token usage, fast startup, and a small linux/amd64 Docker image. The final container is under 50 MB, uses no bundled secrets, supports FIREWORKS_API_KEY, FIREWORKS_BASE_URL, and ALLOWED_MODELS, and has been validated locally with unit, practice, held-out, stress, Docker, and real Fireworks runs.

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.