Lehar Pothugunta@leharpothugunta321757876

Track 1 Agent: A token-efficient, cost-optimized general-purpose agent built for AMD Hackathon ACT II, Track 1. The agent intelligently routes incoming tasks into one of 8 category-specific strategies using local heuristic classification. For each task, it generates an answer using the cheapest model capable of clearing the accuracy gate, leveraging a single tight Fireworks API call. All outputs are validated and coerced locally at zero cost; only when local validation predicts a gate failure does the agent escalate to a stronger model. The architecture combines a local router, per-category prompt/model strategies, an OpenAI-compatible Fireworks client, and a validation layer that ensures output correctness without expensive reprocessing. Includes comprehensive telemetry for token accounting and an evaluation harness for per-category model recommendations. Designed to maximize accuracy while minimizing API spend across a 10-minute execution window with 8-way concurrency.

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.