Lukas OberbichlerSoftware engineer & product builder

Lukas Oberbichler with a city skyline in the background

I turn messy workflows into reliable software.

I build software and AI tools from the first conversation to delivery — as part of your team or as an independent contractor.

For employers
Vehicle software at Bosch, museum installations at Inomedia and planning plug-ins at Build Informed.See my experience
For businesses
I replace copying between spreadsheets and inboxes with software built around how your team works.See the Letter demo

Open to roles and contract projectsRemote, with occasional agreed on‑site collaboration if requested

Main project · in development

Letter

Try an example of your workflow.

Letter is the main product I’m building. Describe how your business works, answer a few questions and try an editable example for enquiries, production or invoicing.

Four checks before a change goes live

  • The goal is written down first
  • Built in a separate copy
  • Reviewed independently
  • Checked against the live version
How I keep Letter reliableTechnical details, the approach behind them and what is still in development.

An approach inspired by the AI Engineering Loop

I’m following the AI Engineering Loop described by Langfuse and others: applications that agents can inspect, evaluate, experiment with and improve. Human feedback defines what “better” means.

How Letter ships changes safely

Agents do much of the building on Letter. Four checks keep every change reviewable and the working version safe.

  1. ScopedAgree what “done” means

    The problem, the result we want and how we’ll check it are written down first.

  2. IsolatedBuild beside the working version

    Changes happen in a separate copy, so the version people use keeps working.

  3. ReviewedA second, independent review

    A separate reviewer checks that exact change and reports any problems.

  4. VerifiedReleased only with evidence

    Release checks tie the reviewed code to what is actually running live.

Pieces I’ve implemented

Observable behavior
Trace export, scores and authenticated readback provide evidence that can be inspected by people and agents.
Explicit change boundaries
Revision-bound briefs, isolated candidates and bounded review-and-repair cycles keep a requested change inspectable.
Checks tied to the result
Release qualification connects the candidate’s source to browser checks and matching observations. Incompatible changes can be held for review.

The AI Engineering Loop

Live activity reveals what needs work. Development turns those observations into tested changes, then deployment starts the next cycle.

  1. TraceOnline

    Capture what the application does in real use.

  2. MonitorOnline

    Find patterns and problems; add human feedback.

  3. Build datasetsOffline

    Turn useful examples into repeatable test cases.

  4. ExperimentOffline

    Try changes to prompts, models and code.

  5. EvaluateOffline

    Compare results against defined quality criteria.

Deploy
Adapted from Langfuse’s AI Engineering Loop. View the original diagram

These are implemented building blocks. The broader dataset, experimentation and evaluation workflow is still being developed.

The Langfuse approach

From car software to AI tools

At Bosch, in museums and in construction software, I’ve worked on the details that turn code into something people can use. Years of direct customer work, including IT support for seniors, taught me to listen carefully and explain technical choices clearly.

Bosch
Car software
Developed vehicle control software and tested how it behaved in prototype vehicles and simulators.
Inomedia
Museum installations
Took over an existing museum application, built Linux video kiosks and installed them on site through to handover.
Build Informed
Construction software
Built plug-ins for Revit, the building-design software, to automate planning work and bring older code up to date.
Tribe
Co-founder
Co-founded a creator-funding product and helped build a music-licensing prototype presented at Red Bull’s London offices. Read the organiser’s account 

What I can help with

How a project starts

For employers: The same first step works for roles — or send me the job description. Email about a role

  1. A 15-minute call

    You describe the workflow or the role. We find out quickly whether I’m the right fit.

  2. A short written plan

    What I’d build first, what it needs from your side and what stays out of scope.

  3. Built in small steps

    You see progress early and can steer it before anything is final.

Let’s work together

Tell me about the role or the workflow you want to improve. Fifteen minutes is enough to see whether we’re a fit.

Book a 15-minute call

Opens Calendly — pick any open slot.

Or write to