I build useful systems at the seam between devices, browsers, backends, and AI.
Backend-heavy engineer · local-first toolmaker · open-source contributor
Distributed systems · Search · Payments · Observability · AI agents · Production LLM infrastructure
I like software that removes a small, stubborn bit of friction—and remains understandable after the clever part is over.
⚡ HotDropLocal, two-way file transfer between macOS and Android. No cloud, cables, or account. Transfers are resumable and protected with local TLS, session tokens, strict cookies, and rate-limited authentication.
|
A private conversion toolkit that runs entirely in the browser. Work with Markdown, PDF, JSON, CSV, images, Base64, and Slack formatting without uploading files to somebody else’s server.
|
|
A tiny native macOS browser picker. Click a link, choose a browser, move on. A focused AppKit utility with no embedded browser engine, network service, telemetry, or Electron-sized RAM tax.
|
🔗 Mac2AndA native clipboard bridge from macOS to Android. Clipboard sync with AES-GCM encryption, Keychain-backed secret rotation, QR onboarding, device controls, and a deliberately documented threat model.
|
The side projects show how I think. My production work shows how far I can take it—from a product surface all the way down through APIs, data systems, infrastructure, and reliability.
| Area | Systems I’ve worked on | Technologies |
|---|---|---|
| Backend platforms | High-throughput APIs, payments, idempotency, reconciliation, and distributed services | Python, Go, Ruby, PostgreSQL, MySQL |
| Search & data | Indexing, analyzers, relevance, low-latency retrieval, and large document collections | Elasticsearch, SQL, Redis |
| Events & experimentation | Kafka pipelines, queues, async workers, A/B testing, segmentation, and rule engines | Kafka, workers, caches, event-driven systems |
| AI infrastructure | Agents, tool-use orchestration, eval pipelines, tracing, debugging, and cost optimization | Python, LLM SDKs, Sentry, Grafana |
| Cloud & reliability | Kubernetes platforms, AWS networking, load balancers, incidents, and performance tuning | AWS, EKS, Kubernetes, Docker, Terraform |
| Product engineering | Browser tools, native macOS utilities, and full-stack product surfaces | TypeScript, React, Swift, AppKit |
| 200M+ users served |
200M+ events / day |
10M+ search documents |
99.9999% payments uptime |
~22% lower LLM cost |
That range is deliberate: I’m comfortable moving between product decisions, application code, distributed architecture, infrastructure, and the ugly production edge cases in between.
| Principle | What it means in practice | |
|---|---|---|
| 🎯 | Useful over flashy | Start with real friction and make the happy path obvious. |
| 🔒 | Local-first when possible | Keep private data on the user’s device and remove unnecessary infrastructure. |
| 🧱 | Simple surface, serious internals | Small tools still deserve resilient protocols, tests, and explicit security boundaries. |
I contribute across AI infrastructure, SDKs, developer tooling, testing, and backend reliability:
Langfuse · pytest · LangChain Google · LiteLLM · MCP Toolbox · Crush
Working set: Python · Go · Ruby · TypeScript · Swift · SQL · Kafka · Redis · PostgreSQL · Elasticsearch · Kubernetes · AWS · Docker · Terraform · LLM tooling
Build small. Think in systems. Ship the useful thing.
Portfolio
·
Projects
·
Open source