At OxiqAI, I built a real-time AI meeting platform for live communication, speech transcription, and synchronized meeting insights. I also engineered a low-latency pipeline for speaker-aware transcripts using voice activity detection, diarization, and transcript synchronization.
I built an agentic Master–Sub-Agent architecture for enterprise workflow automation, with task decomposition, intelligent delegation, MCP-based tool calling, and multi-agent execution. I coordinated team development by tracking progress, reviewing implementations, resolving blockers, and delivering assigned modules.
In my projects, I developed machine-learning systems for credit card default risk prediction and image–idiom alignment. I also built and deployed a GitHub RAG assistant with hybrid retrieval and an LLM-based critic agent.
