Mubtasim Fuad@tryhard_fuad

Taskly is a Dockerized Python agent built for the AMD Developer Hackathon that solves a core challenge in general-purpose AI agents: no single model or prompt style is optimal across every task type. Taskly reads incoming tasks, classifies each into one of eight categories — factual QA, math reasoning, sentiment classification, summarization, named entity recognition, code debugging, logic puzzles, and code generation — using a lightweight keyword-based classifier with zero added token cost. Each category is routed to a category-tuned prompt template and the most suitable Fireworks AI model from the allowed set, with code-related tasks routed to a code-specialized model. Templates are deliberately terse and forbid markdown formatting to minimize output tokens, since scoring rewards accuracy first and token efficiency second. The pipeline includes per-task error handling so a single failed API call never crashes the full run, ensuring reliable, valid JSON output every time within the runtime and size constraints.

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.