ZeroLatency
Team consisting of a Centific Full Stack Developer skilled in Python, asynchronous LLM orchestration, voice AI (Deepgram, ElevenLabs), and Twilio API integrations.
Project Description
Latency Zero is a local-first voice/action agent built with Python, Streamlit, Cactus Compute, and FunctionGemma.
It handles most requests on-device for fast response, privacy, and high edge usage.
A hybrid router checks confidence, tool-call validity, and query complexity to decide when to escalate to Gemini.
This keeps simple tasks local and uses cloud only for harder or ambiguous multi-tool requests.
Agentic workflows run through structured tool schemas, argument validation, and execution tracing.
Overall, it combines speed, correctness, and intelligent edge-cloud routing for strong hackathon judging criteria