Team Idea
We are building an enterprise-grade, self-correcting Agentic Pipeline powered by the AMD MI300X. Instead of just generating a static synthetic dataset of Flutter code, our pipeline uses a multi-agent architecture (Planner/Coder) with an automated "Critic" loop (running flutter analyze and flutter test via CLI).
The ultimate goal is to generate high-quality "Agentic Trajectories" (Chain-of-Thought + CLI tool usage + Error Correction) to fine-tune smaller open-weight models (like Qwen) into autonomous Senior Flutter Developers. We are utilizing Evol-Instruct methods to systematically multiply app complexity before generation, ensuring our fine-tuning dataset covers extreme edge cases.
Who we need: We are looking for AI Engineers with experience in vLLM deployment, dataset curation (JSONL ChatML formatting), or Flutter developers who want to help refine the automated testing architecture.
Team Members
Aiman Alshalfi
abo00351631623
Submission
Team Leader hasn't made a submission yet