.        .              .
        retrieval ----> memory ----> action ----> feedback
              \             \            \             \
               \             \            \             \
                +---- vision --+-- tools --+-- systems --+

        I like the space where models stop being demos
        and start becoming instruments people can actually use.

I am a machine learning engineer focused on the line between research and usable software: RAG systems, agentic workflows, computer vision, and backend infrastructure that can survive real users.

At iCog Labs, I work on context-aware retrieval for MeTTa code, query rewriting, RAG evaluation, secure API-key handling, and production pipelines around LLM systems.

• Make AI systems observable, testable, and calm under weird inputs.
• Give models better context instead of asking them to guess harder.
• Build interfaces where automation feels understandable, not magical.
• Treat research prototypes as seeds, then grow the engineering around them.

• Built retrieval and evaluation workflows for RAG systems over symbolic code.
• Solved 400+ LeetCode and Codeforces problems through Africa to Silicon Valley.
• Won Reboot The Earth with AgriLo, an AI assistant for smallholder farmers.
• Led a Top 3 global team at OpenEPI x UNLEASH and won the Grand Finale in Kigali.
• Scored Top 100 out of 4000+ submissions at the IBM TechXchange Watsonx Hackathon.

Great things start from nothing.