At Lythe Labs, I contributed to three open-source AI agent projects and built an Agent Testing Suite dashboard for latency, API cost, and tool-call frequency. I also designed a Voice Agent Starter Kit and engineered a long-running agent with persistent memory and resumable sessions.
At School of AI, I built domain-specific AI systems, including a medical note structurer, an earnings call analyzer, and an incident response assistant. At AABS, I developed RAG pipelines and PDF chatbots, improving retrieval precision through embedding and relevance tuning.
I’ve contributed to Arize-ai/phoenix and docling-project/docling-core, and built projects including Crucible, Relay, and ai-cache. My work spans LLM applications, multi-agent systems, retrieval, and machine learning.

