About
Solo engineer, real products.
I'm an AI software engineer — shipping software since 2022 and focused on production AI since 2024: GenAI/LLM applications, agentic and RAG systems, and predictive ML. I work across the stack: FastAPI, Next.js, Postgres, and Docker, with auth, CI/CD, and security hardening baked in.
AI/ML Engineer at Sevina Technologies · healthcare-AI automation, by day
By day I build healthcare-AI automation at Sevina Technologies — clinical-compliance and reimbursement pipelines (constrained by HIPAA, so shown here only in the abstract). On my own time I ship live AI products on a $0 free-tier stack, which forces discipline: no waste, real engineering, shipped.
The one rule across all of it: every number a user sees is genuinely computed — never faked.
How I think
Three rules I don't break
- Evidence over demos
- Every number is genuinely computed. If a number can’t be shared, I describe the mechanism and say so — I never invent one.
- Safety is structural, not statistical
- When an action can do damage, I put a deterministic gate in front of it that the model cannot talk its way past — and fail closed, refusing when safety can’t be proven.
- Constraints force good engineering
- A $0 stack rules out waste: retrieval-scoped prompts, bounded agent loops, lean containers. The limits shaped the design from the start.