Salmon Ouma

I build and evaluate AI systems that hold up in production. With 5+ years as a software engineer, I’ve gone from scaling secure sign-in for 2M+ customers to building LLM systems end to end: a production RAG pipeline, an MCP-based agent, and a WhatsApp AI assistant, as the sole engineer at XiOps. I’ve also worked as a contract calibrator and senior Python engineer, where I led a team creating adversarial coding scenarios that expose LLM weaknesses and reproduced real GitHub issues (SWE-bench style) for model training and evaluation. Today I’m Tech Lead at Regional Business Connection, taking AI-powered products (computer-vision surveillance and smart parking) from concept to deployment with Django, Kafka/Spark, and OpenCV.
What I bring
• LLM evaluation: benchmark and task design, failure analysis, reference solutions
• Applied AI engineering: RAG, pgvector, agents/MCP, async pipelines
• Production foundations: Python/Django, React/TypeScript, AWS, Terraform, Docker
• Research grounding: published work on mathematical modelling for policy making.
Open to research engineering, LLM evaluation, and applied AI roles