Software engineering
powered by AI.

As soon as possible — typically within 1–2 days of our first call, including scoping and setup. I keep capacity reserved for new engagements so you're never stuck waiting weeks to start.

Full stack across web and mobile — React, Next.js, TypeScript, Node.js, Python, Swift, Kotlin, React Native, and Flutter — plus cloud (AWS, Google Cloud, Firebase) and databases (PostgreSQL, MongoDB, Redis).

I also specialise in AI orchestration and automation: LLMs, RAG pipelines, multi-agent systems, vector databases, and production-grade LLM integrations.

Both. I've shipped products for pre-seed founders with nothing but a Figma, and for established companies who needed a senior hand on a focused initiative. The engineering rigor doesn't change — only the pace and scope.

Yes — it's a specialty. I've shipped end-to-end encrypted products where the server is deliberately blind to user data, including Flowvault, an open-source zero-knowledge encrypted notepad with plausible deniability, drand-backed time-locked notes, self-destructing Encrypted Send links, and a dead-man's switch.

In practice that means browser-side cryptography (Argon2id for password-to-key derivation, AES-256-GCM for authenticated encryption, WebCrypto throughout), hidden-volume formats, auditable Firestore security rules, and a clear, published threat model. If you're building something where user data must never be readable by your own servers — password managers, encrypted notes, secure messaging, health or legal tools — I can take it from threat model through production.

Two ways. First, as a daily tool — I use it constantly to scaffold code, rewrite things, and sanity-check my thinking. It makes me faster, but I don't trust its output blindly, especially on anything security-sensitive.

Second, and more interesting, I build AI into the products themselves — RAG systems, agents, LLM features shipped to real users. The hard part there is almost never picking a model. It's the boring stuff around it: keeping costs sane, catching hallucinations, handling failures, and making sure the thing doesn't fall over at 3am.

That's where most LLM prototypes die on the way to production — so that's where I spend my time.

Depends on scope. Most full-build engagements run 3–6 months end-to-end. Smaller integrations, MVPs, and focused feature work can ship in weeks.

Yes — happy to sign an NDA before any detailed scoping discussion. Just send your template and I'll review and return it the same day.

Both work. Hourly is $40–$80 depending on scope and technical depth. Fixed-price is quoted after a short discovery so the number reflects the actual work, not a guess.

Yes. Launch is rarely the end — I offer ongoing maintenance and enhancement retainers so production issues, feature requests, and upgrades are handled by someone who already knows the codebase.