Backend engineer who builds systems where being wrong isn't an option — then points that discipline at AI, and translates it for the people who have to trust it.
I work on a financial core serving 100M+ accounts, where a mismatched record isn't a bug — it's an audit finding.
90M+ account records migrated & reconciled @ 99.9%+ validated accuracy
700M+ domain records owned end-to-end · published to 4+ downstream services
60–70% throughput gained under a 2–3M event/day workload
Correctness under load — I make distributed systems provably correct: migration, reconciliation, event ordering, and the race conditions that silently mislink data if you let them.
AI reliability — I build open-source tooling that answers the question a demo can't: does your agent actually hold up in production?
Translation — Correctness only counts if the people who don't write code trust it. I defend system behavior to compliance, audit, and tech-lead reviewers. It's the part I like most.
Open-source benchmarking for tool-using LLM agents. Plug in your agent, build evals from a template, and benchmark it on reliability, latency, and cost — with a full trace behind every run.
- 12-case benchmark porting real distributed-systems failures — duplicate-key retries, DLQ poison messages, config drift, rate-limit storms, unsafe remediation
- Typed Agent-Under-Test contract + HTTP JSON adapter — register any external agent against the same suite as the built-in reference agent
- FastAPI · PostgreSQL · Pydantic · SQLAlchemy · Alembic — runs, trace persistence, eval execution, comparison APIs
- MIT licensed, with methodology, architecture docs, and tests
Java · Python · SQL · TypeScript
Spring Boot · FastAPI · Kafka · PostgreSQL · AWS · Docker
Event-driven architecture · idempotency · retry/DLQ handling · eventual consistency · reconciliation · LLM evals · agent tracing · guardrails · structured outputs
Forward Deployed / AI Solutions Engineering — building frontier-model systems in production and explaining them to the people who have to trust them.