Voice Agents Hackathon
The era of "AI Demos" is over. The next frontier is Production grade AI agents. Cekura and Daily, in partnership with NVIDIA and AWS, are challenging builders to bridge the gap between a Voice Agent that "works" and an agent that scales, persists, and learns. This hackathon is designed for engineers who want to dive deep into the infrastructure and evaluation loops required to run state-of-the-art voice agents in the real world.
About
Key Themes: Leveraging and customizing SOTA open weights models Infrastructure and network optimization Auto-Improvement harness We aren't just looking for the best-sounding voice; we are looking for the best system. Your challenge is to build a voice agent that utilizes a continuous feedback loop: Build & Customize: Leverage NVIDIA-accelerated SOTA open-weights models and AWS compute to create high-reasoning voice engines. Deploy at Scale: Use Pipecat’s high-performance infrastructure to optimize network performance and eliminate latency. Simulate & Evaluate: Implement Cekura’s automated testing and evaluation platform to move beyond "vibes." Auto-Improve: Create a system where evaluation data flows back into the agent to improve performance, reliability, and accuracy over time. This Hackathon is for developers interested in building voice ai agents. Hackathon Schedule: 8:00 AM – Doors open & registration 8:30 AM – Breakfast 🧇 9:00 AM – Welcome / Hackathon begins 12:00 PM – Lunch 🥪 6:00 PM – Submissions due 6:00 - 8:00 PM – Dinner & conversations 7:00 PM – Finalist Demos 8:30 PM – Winner announcement 🏆 9:00 PM – Closing Prizes Guaranteed Interview with YC Swag by Cekura and Daily Special judges’ prizes from NVIDIA and AWS
This hackathon has ended