Oskar Freye.

Software Engineer · AI Operations · Technical Leadership

I ship AI systems and rebuild broken stacks.

Technical founder, 14 years shipping code across the stack. Most of my work is hands-on with founders and small teams, usually writing code alongside them. Agents running in production. Code that still makes sense in six months.

14 yrs Shipping code
Startup CTO
Live AI in prod
Download CV
Current stack
TypeScript Go Elixir Python AWS · SST DynamoDB Three.js Next.js Astro React gRPC Docker Claude · GPT LangChain EventBridge PostgreSQL TypeScript Go Elixir Python AWS · SST DynamoDB Three.js Next.js Astro React gRPC Docker Claude · GPT LangChain EventBridge PostgreSQL
01

AI Operations Layer

I build autonomous systems for the processes you'd rather not touch again. Humans stay on the calls that matter. Not chatbots — agents that read real context and trigger real actions.

Learn more →

Tech Turnaround

Codebase no one wants to touch? Team walked out? I come in, figure out what's salvageable, stop the worst bleeding, and rebuild the rest. After 14 years you develop a nose for where things actually break.

Technical Leadership

Retainer work for teams that want a senior voice in the room without hiring another CTO. Architecture calls, code review, AI strategy. I'll still open a PR when writing the code is the fastest way to answer the question.

02

14 Years Across the Stack

Started coding at 12. Since then: WebXR experiments, Laravel backends carrying too much history, serverless SaaS, internal tooling, infra I've had to wake up to at 3am. Shipped at every layer of the stack. IHK-certified Fachinformatiker, if that matters to you.

Startup CTO — Built & Shipped

Several products from zero to real users. A few that nobody ended up using. Both teach you things, mostly which corners are safe to cut and how to keep shipping while the spec is still moving.

Cloud Native — Serverless & Containerized

Serverless-first on AWS: SST, Lambda, DynamoDB, EventBridge. Containers when they're the right tool, not by default. TypeScript and Go for most of what I write. Elixir when the problem fits. Infrastructure lives in code, not in someone's head.

AI & Agent Systems in Production

I'm past the demo stage. Multi-agent orchestration, custom tools wired into real APIs, guardrails that catch things before they hit prod. Agents run in my own business daily — lead gen, code review, the boring parts of operations. If one dies, I know about it within the hour.

03

Got a system that needs to actually work?

Rescue job, greenfield AI build, or fractional CTO engagement. I read every message and reply inside a day.

Typical reply: < 24h · Based in Germany · Working worldwide