At Lume AI, I built an MCP-based low-code orchestration platform for LLM agents to run API integrations, browser automation, and custom tools. I also developed an integration framework supporting 33+ external services, with production monitoring and agent tracing.
At TransHumanity, I designed a Python ingestion platform processing 1M+ real-time events weekly and built RAG pipelines using OpenAI embeddings and FAISS. I developed microservices and multi-agent workflows for TraffEase, a Manchester Prize finalist platform.
At ACTO, I designed Python- and SQL-based data pipelines that automated reporting workflows, reducing recurring reporting effort by 100+ hours per month. I also developed XGBoost models for upsell forecasting and customer segmentation, contributing to an 8% increase in conversion rates.

