Case Studies
Deep-dives into hard problems. Implementation secrets omitted by design — the judgment is the point.
Beating the Pirates: A Custom Content-Protection Protocol for a Closed Video Platform
Off-the-shelf DRM failed and cost too much. So I designed a custom content-protection protocol — on a standard-cryptography foundation — for a closed video platform. 20,000+ users, 5–10k concurrent, still holds today.
Read case study →We Didn't Need a Message Queue — We Needed an Event Bus
The 'microservices' were really a distributed monolith talking over RPC — and when a multi-service call failed halfway, the code manually rolled back the earlier calls. What if the rollback failed? No answer. So I moved the platform to a Kafka event bus.
Read case study →How Do You Test an AI Agent? I Open-Sourced a Library for It
An AI agent's reply is different every run, so assert.Equal is useless — but most eval frameworks over-correct and LLM-judge everything. I built a narrow library that judges only what a plain if can't, and open-sourced it (zero dependencies, bring-your-own-model).
Read case study →A Production AI Agent That Runs on $28 a Month
A full multi-tenant SaaS that plugs into businesses' Meta channels — WhatsApp, Messenger, Instagram — so their customers chat with an AI that quotes and books. Everyone builds these in Python with LangGraph; for a low-end box that's bloat. I built it in Go, and it runs the whole thing for about $28 a month.
Read case study →Building Go's Goroutine Scheduler From Scratch in C
I use goroutines every day but couldn't explain what one really is. So I rebuilt the core of Go's runtime — stackful coroutines on an epoll event loop — from scratch in C, and proved three concurrent I/O tasks overlapping on a single thread.
Read case study →About
I'm a senior backend engineer with a Solutions Architect background and hands-on production AI/LLM experience. I've built the hard parts of modern backends — event sourcing, CQRS, the transactional outbox pattern, Kafka pipelines, CDC, and durable workflow orchestration — and shipped them across microservice platforms.
Earlier I was the founding engineer and Solutions Architect of a secure video LMS, scaling it to tens of thousands of users. My range runs frontend / mobile / desktop → architecture → distributed backend → AI, but backend and distributed systems are where I do my deepest work.
I care about systems that are correct under load and observable when they aren't. I'm open to senior backend roles — Go, Node.js, or a stack worth learning. Remote-friendly.
Core
- Distributed systems & event-driven architecture
- Microservices · DDD · CQRS · Event Sourcing
- Apache Kafka · outbox · CDC
- PostgreSQL · Neo4j · Redis
- AI/LLM · RAG · tool-calling agents
- Docker · Kubernetes · OpenTelemetry
Get in touch
Open to senior backend / AI-backend roles. Remote-friendly.