Based in Germany · Remote across DACH, US & UK
I help businesses automate processes and scale with bespoke AI-powered software.
Building modern, clean and maintainable production-ready software, tools and AI applications that are delightfully user-friendly, reliable and robust, using quality gates, observability, testing and evaluations.
Learn moreWorked with
Most businesses don't realize that a lot of manual processes between tools can be automated and scaled without a human admin, which is also more fault-tolerant and time-efficient.
- Step 1
Audit processes
I speak with the business owner, uncover the real problems, and evaluate the processes: what can be automated, and where would AI-powered software help the business?
- Step 2
Design the solution
I design and propose a solution with state-of-the-art technology, and excellent user experience.
- Step 3
Test it with users
The pilot is tested with real users through fast iterations, uncovering their needs and workflows.
- Step 4
Optimize and run it in the real world
An evaluation framework is set up when needed and the software is optimized against it. Quality gates, observability, testing and evaluations are all in place, running on production servers, durably and at scale.
- Step 5
Hand off with proper documentation
The whole system is handed off with proper documentation and a clean, maintainable codebase, ready for third parties to use.
Case study
Research Data Platform
Maia Earth Organization
In production with 56+ participants and 74+ sessions recorded, physiological data is captured and assessed with no clinical setup. Every session becomes research-grade evidence toward non-invasive, sound-based protocols any clinic can run.
Built end to end, working with the Data and ML architect and researchers, from the web app to the backend and production infrastructure, with the data pipeline integrated so every session is analyzed automatically. It stays reliable in production, so researchers, participants and clinic staff run it without help and the research data holds up.
The Data and ML architect's validated HRV analysis (RMSSD, HR, DFA) went straight from research notebooks into the platform as one source of truth, so every session report runs the exact analysis they validated.
Data collection used to mean manual steps, spreadsheets, and one person holding it together. Now participants and researchers use the platform directly, and every session runs through the data pipeline automatically, analyzed and turned into a finished report.
Case study
Knowledge Retrieval System
Open-source project
Processes 200+ source documents into one searchable knowledge base. Questions that once took days or weeks of reviewing source files are answered in seconds, with the right context surfaced and every answer grounded in citations. A built-in Q&A memory keeps past answers at hand and avoids duplicate questions.
Designed and built a complete RAG pipeline end to end, from multi-format ingestion (ASR transcripts, PDFs, EPUB) and token-based chunking to embeddings in PostgreSQL/pgvector, a FastAPI service and a Next.js frontend.
Implemented hybrid retrieval, semantic and keyword search fused with Reciprocal Rank Fusion plus reranking, exposed as tools to a Pydantic-AI agent that streams answers grounded in cited source passages, so a confidently wrong or unsourced answer never slips through.
Built an evaluation framework with LLM-as-judge scoring, a custom error taxonomy and Langfuse observability, so answer quality is measured and improved iteratively, not guessed.
Case study
Enterprise Data Hub
openpack GmbH
Resulted in clean, maintainable code, a polished user interface and user experience, and performance and reliability for industry-wide data exchange.
Developed and maintained the React/TypeScript frontend of the Enterprise Data Hub, onboarding quickly into an existing codebase and delivering features independently, from data visualisations built on the backend data models to an AI-assisted translation feature.
Worked in sprints with the product owner, tech leads, UX and backend, turning UI/UX designs into precise interfaces and writing the frontend-facing specs for the backend APIs and the Auth0/Keycloak integration.
Kept code quality high by refactoring existing code alongside feature work, and through reviews, quality gates and CI/CD, with Playwright end-to-end tests, Biome and SonarQube.
About
I bridge data, design and engineering.
AI products break in the gaps between data, design and engineering. I don't replace the Data and ML architect or the designer, I speak all three languages and implement the whole thing, so their work ships as one product instead of falling through the handoffs.
- 4.5 years of production experience, including in teams.
- Ships complete projects solo, production-ready end to end.
- Certified in UX design, focused on the experience.
- Works natively with AI, with the right quality gates.
- Experienced in AI evaluations.
- On the frontier of where AI actually applies.
I'm based in Bad Lippspringe, a small Kurstadt in Germany, and work globally with clients across the DACH region, the US and the UK.
Testimonials
“Hac Hai has a very effective work methodology, both collaboratively and async.”
“I worked with Hac Hai, building a Data Platform for HRV biofeedback and sound data which has evolved from Prototype to Productive use. It is now successfully used by medical staff members and session facilitators in 2 clinics. It has wide community support and interest to adopt solution, and attracted interest from field experts (neuroscientists). He is reliable and always delivered a robust solution timely. I was impressed by how elegant and simple the solutions he built were, both solving users needs as well as minimising the development and maintenance overhead.”
Contact
Have a process to automate, or an AI feature to ship?
Whether it's manual work between tools that should run itself, or an AI product that needs to reach production reliably, I'll take it from audit to a robust, well-tested build, end to end or alongside your team.