There's a hackathon almost every week right now, agentic AI, open source, AI infra, all of it. I want to enter more of them, and I want to build the team differently than most people do.
What I bring
I'm a computer science undergrad working as an AI Systems Engineer Intern at RYSE Technologies, where I built the main orchestrator agent that routes tasks across a registry of 3,000+ specialized subagents through intent classification, capability scoring, context assembly, and dispatch. That cut misrouted tasks by roughly 40% and brought dispatch latency to around 120ms.
Outside of that, I've built three agentic AI systems, all open source:
RabbitHole, a multi-agent "courtroom" system on LangGraph that runs adversarial debate personas. I fixed a token-budget overrun where agents were blowing 3 to 4x past budget per round with no shared memory, just prompts. Moving the constraint into a structured state schema cut latency 50% (19.8s to 9.8s) and input tokens 70%, measured in LangSmith.






