Lukas Möller

Senior Product Engineer

I turn unclear product ideas into software that holds up in practice.

About

Hi, I’m Lukas. I take products from the first idea to production.

Portrait of Lukas Möller

I’m based in Berlin and have worked across backend systems, full-stack product interfaces and infrastructure. I like staying close to both the product decisions and the implementation.

Right now, I’m building Phemilo, a native strength-training product for iPhone and Apple Watch. Alongside it, I work on retrieval, AI evaluation, internal tools and integrations.

Building Phemilo Available for permanent roles and freelance projects.

Work and notes

One product I build and operate, followed by two notes on the technical questions behind my recent AI work.

Independent product · iOS · AI fitness

Phemilo

A native iOS and watchOS strength-training product for progressive overload, training history and AI-supported coaching. I designed, built and released it independently.

I own the native apps, backend, subscriptions and release. The AI coaching work includes workout-history analysis, plan generation, retrieval, tracing and evaluation.

  • SwiftUI
  • Supabase
  • Langfuse
  • RevenueCat
Phemilo analytics on iPhoneLogging a set with Phemilo on Apple Watch

02

Experience note · Retrieval and citations

Document Q&A That Can Say “I Don’t Know”

What I learned about keeping document answers grounded, traceable and willing to stop when the evidence is not enough.

Problems I worked through

  • Keeping answers connected to evidence
  • Knowing when the system should refuse
  • Testing retrieval separately from generation
Read what I learned

03

Experience note · AI evaluation and observability

How I Evaluate AI Features Before They Reach Users

The evaluation habits I use to make prompt, model and tool changes without relying on a convincing demo and intuition alone.

Problems I worked through

  • Turning failures into reference cases
  • Comparing prompt and model changes
  • Deciding when automated evaluation helps
Read what I learned

A few habits I rely on when the path is not obvious.

  1. 01

    Start with the smallest useful version

    I narrow the problem first: who needs what, what can wait and what the first version has to prove.

  2. 02

    Own what happens after release

    Architecture matters, but so do tests, deployment, monitoring and the next iteration. I prefer to stay close to all of them.

  3. 03

    Make AI failures visible

    I keep retrieval, prompts, tool calls and evaluation observable so failures become cases to fix rather than anecdotes.

More than seven years building software across product companies, agencies and independent work.

06.2025—NOW

Independent Product Engineering

Senior Product Engineer

Building Phemilo’s native apps and AI coaching workflows, including workout-history analysis, plan generation, retrieval and evaluation.

11.2022—06.2025

Productsup GmbH

Backend Developer (Full Stack)

Optimised Symfony and Node.js services, improved API response times by 35% and built an Angular activity log for enterprise user actions.

11.2020—10.2022

Epikur Software GmbH

Software Developer

Led work on Angular modules for a healthcare platform, expanded unit and integration tests and supported reviews, pairing and junior developers.

03.2020—11.2020

VISC Media

Full-Stack Developer

Built Visticle, a multi-tenant Laravel commerce platform that synchronised products with Amazon, Zalando and other sales channels.

02.2019—03.2020

Independent freelance development

Freelance Developer

Prototyped a multi-tenant commerce platform, built a Flutter application and handled backend migrations and maintenance for clients.

06.2018—02.2019

fotograf.de

Junior Software Engineer

Worked on a large e-commerce platform built with CakePHP and AWS-based microservices.

If the work sounds like a fit, let’s talk.

I’m open to permanent roles and focused freelance projects.

For teams Senior product, backend or full-stack engineering roles.

For projects Product builds, integrations and AI evaluation work.