Yosevu Kilonzo
pragmatics · fullstack · context engineering for agentic systems

Hi, I'm Yosevu. Senior Full Stack Engineer with a linguistics background. I think carefully about how language models interpret context, structure output, and fail.
I build AI agent systems and the infrastructure that makes them safe to use: LLM orchestration, tool use, structured outputs, and grounded generation.
How I work
I work across the full stack (frontend, APIs, infrastructure) and reach for AI where it earns its place: retrieval over fine-tuning, structured outputs over free-form, the right model for the task. I communicate tradeoffs clearly across product, design, and engineering.
What I care about
- Grounded AI systems: evaluation, retrieval quality, safety by design. I build AI features that cite sources, resist hallucination, and degrade gracefully.
- Judgment over process: knowing what to build and what to skip. Embeddings over fine-tuning, single-turn over conversation, cosine similarity over keyword search.
- Product partnership: tight loops with PMs, designers, and stakeholders so technical work serves real user outcomes, not engineering for its own sake.
- Creativity across disciplines: connecting ideas across linguistics, engineering, and design to find non-obvious solutions.