Michael Hausegger
Freelance · Remote
Portrait of Michael Hausegger

Michael Hausegger

Senior Solutions Architect for enterprise software, cloud-native platforms, and RL environments for LLM post-training.

I've spent 20+ years delivering enterprise software end to end: Java and Oracle performance, Kubernetes and OpenShift, security and compliance. I work out what the customer actually needs, then build it in their stack. I now also build verifiable-reward RL environments that train coding agents on real database performance work.

See the RL environment Get in touch
Graz, Austria Remote, 4+ hrs daily overlap with US Eastern Time English · German
  • 20+ years of enterprise softwareArchitecture, delivery and technical leadership
  • Red Hat-certified Helm chartDynaCov Safety Automation, OpenShift ecosystem
  • Merged into Guava, Commons, Jetty, MavenOpen-source contributor across the Java ecosystem
  • MSc Advanced Security EngineeringBuilt an ISMS aligned with ISO 27001
RL Environments · License or custom build

Did the agent make a real database cheaper to run?

Most RL environments for coding agents check whether the agent produced the right answer. Mine check something harder. The agent gets a slow SQL query, its own copy of the database and a shell, and has to tune the query.

It is scored on logical reads: how many database pages the query really touches. Not wall-clock time, which mostly measures the machine. Not how clean the execution plan looks, which is what the engine intends to do, not what it does.

The task sits where an environment is useful for training: frontier models succeed sometimes, but not reliably.

Explore the environment Request a demo under NDA

Can a model tune a slow SQL query all the way to the reference solution?

Scored on logical reads. Original query: 11,238 pages. Reference solution: 115 pages.

  • Claude Opus 5: 3 of 5 runs fully solved, mean reward 0.69
  • Claude Fable 5: 2 of 5 runs fully solved, mean reward 0.89
  • Gemini 3.6 Flash: 1 of 5 runs fully solved, mean reward 0.86
  • GLM 5.2: 1 of 5 runs fully solved, mean reward 0.87
  • DeepSeek V4 Flash: 0 of 5 runs fully solved, mean reward 0.41
One dot per run, 5 runs per model. DeepSeek V4 Flash: first 5 of 10 runs shown; none of the 10 fully solved it. Michael Hausegger · SQL query-optimization RL environment.

Graded on what the database did

Logical reads are what a DBA tunes for on a shared server. In one batch, the two "obvious" join indexes made the query almost 4× more expensive than adding no indexes at all, while the execution plan looked better.

A gradient, not a cliff

The reward is monotonic: every page saved raises the score, so a model gets a learning signal instead of a pass/fail cliff. Scores spread from 0.24 to 1.0, and the same submission always gets the same score.

Hard to game

Grading happens in a clean room on a fresh copy of the database, so shortcuts that lower the page count without tuning anything don't pay. Every rule the prompt states, the grader enforces.

Built on Prime Intellect's verifiers framework, running in a pinned, offline-capable Docker image. SQLite keeps the demo small and self-contained, but nothing in the approach is SQLite-specific. PostgreSQL and Oracle are next.

Freelance services

What I can do for your team

Contract and project work, remote from Austria with daily overlap with US Eastern Time. I start from the problem, then pick the technology.

Solution architecture

Enterprise systems from requirements to production, primarily in Java/Spring and Java EE, with Go or Python where they fit. I also take technical leadership of the delivery.

Java · Spring · Java EE · API design · Microservices

Database & performance

Oracle, PL/SQL and SQL tuning. I've taken queries and programs from minutes to seconds, and migrated legacy Oracle Forms to modern web stacks.

Oracle · PL/SQL · SQL tuning · PostgreSQL · SQLite

Cloud-native & DevOps

Software built for Kubernetes and OpenShift, including a Red Hat-certified resilience-testing product. CI/CD on Jenkins and CircleCI, deployments on GCP.

GCP · Kubernetes · OpenShift · Istio · Prometheus · Grafana · Jenkins · CircleCI

Security & compliance

Secure SDLC, code reviews and penetration testing, backed by six years as ISMS lead for the university management system developed at Graz University of Technology, and an MSc in security engineering.

ISO 27001 · Secure SDLC · Pentesting · Burp Suite · Kali

Track record

Experience

  1. 2026 – present

    RL Environment & Eval Engineer (LLM post-training)

    Independent
    • SQL query-optimization RL environment on Prime Intellect verifiers. The reward comes from SQLite logical reads, correctness is gated by comparing result sets, and every rollout runs in its own Docker container.
    • Calibrated to discriminate rather than saturate: on the hard task, frontier models fully solve 1–3 of 5 runs.
  2. 2022 – present

    Founder & Principal Technical Architect

    DynaCov Inc. · US start-up, Google for Startups member
    • Designed and built DynaCov Safety Automation, a resilience-testing platform for Kubernetes and OpenShift aimed at banking and finance. Go and Istio, with verdicts evaluated against Prometheus metrics. Helm chart and container image certified by Red Hat.
    • Re-engineered the EvoSuite test generator to run distributed on GCP (124-core VMs, Pub/Sub) and feed results into a custom IntelliJ plugin.
    • Owned the whole technical foundation: cloud, CI/CD, DevOps and code architecture, working with co-founders in Silicon Valley and Switzerland.
  3. 2018 – 2022

    Senior Software Architect & DevOps Lead

    CodeLabs GmbH · client: a Swiss consulting company
    • Owned the architecture of a new Java/Vaadin test-automation framework for testing Avaloq banking software, and set its technical direction.
    • Built the CI/CD pipelines (Jenkins, later CircleCI) and ran deployments on Google Cloud Platform.
  4. 2002 – 2017

    Graz University of Technology

    Software development department, building the university's commercial management system
    • Roles over 15 years: Oracle expert → team lead & technical product manager → IT security expert & ISMS lead → test automation.
    • Built core Oracle and PL/SQL applications for a commercial university management system, and tuned their performance.
    • Led a development team and owned a product segment, including customer contact, training and hiring.
    • Built an ISMS aligned with ISO 27001, ran security audits and penetration tests, and served on the university's Data Protection Advisory Board.
Open source

Code merged into projects you depend on

Bug fixes, performance improvements and test coverage across major Java ecosystems. More on GitHub.

Contact

Let's talk

Need RL environments for database or performance engineering, an architect for a hard problem, or a slow database fixed? Email is the fastest way to reach me. I usually reply within a day.

Email me

Also open to full-time remote roles with US teams.