Senior software engineer
Ten years architecting and scaling platforms for millions of users — from modularising legacy monoliths to designing multi-agent AI systems. Leading squads of 5–8 engineers, owning it end-to-end, and taking it from messy to maintainable.
- Production LLM serving — vLLM on Cloud Run GPUs, OpenAI-compatible endpoints, full Terraform
- Fine-tuning pipelines — LoRA on compact models, synthetic QA generation, orchestrated end-to-end with Metaflow
- Performance engineering — Rust N-API addons that 3× Node.js throughput; WASM when it fits
- Agentic tooling — MCP servers, RAG + knowledge graphs, multi-agent orchestration
- Full-stack delivery — Schema to pixel: REST/GraphQL APIs, strict data contracts, native & web clients
| Project | What I did |
|---|---|
| Mission-critical pharma intelligence platform | Modularised 40% of a legacy monolith via Strangler Fig, then delivered a greenfield v2 in under a year. Owned full-stack delivery (native / .NET). |
| Global live-streaming & OTT platforms | Led squads delivering sub-5s live latency at 99.9%+ uptime. Built a custom tvOS playback engine and Server-Driven UI, cutting new screen launches from weeks to minutes. |
| Hyper-scale streaming app | Served millions of users at 60fps on low-end hardware by migrating a hybrid app to 100% native. Built a custom Gherkin→UI test framework on remote device farms. |
- AI / ML: Python, LangChain, RAG, Fine-tuning, Agentic Systems
- Full-stack: TypeScript, Node, React, Flutter
- Infra / DevOps: Docker, Terraform, CI/CD, AWS / GCP
- Performance: Rust (N-API, WASM)
- Native mobile: Swift, Objective-C
- Preference on FullStack roles
Building something ambitious? Let's talk. Prague, CZ · boris.raptor@gmail.com · LinkedIn · Blog