Consultancy · AI & Data Engineering · Munich
We build the automation and AI that removes your manual work.
SAP and PLM migrations, messy master data, offline AI. Data-heavy work for teams where correctness matters.
Automation, AI, and integrations.
AI & LLM integrations
Extraction, cleaning, agents, and fine-tuned models inside your product or process. Cloud, local, or fully offline.
Automation & internal tools
CLIs, bots, and small services that eliminate repetitive work.
Data mining, integrations & pipelines
Collectors, APIs, and pipelines that move data between systems.
Open-source projects.
Client work is under NDA, but these repos show how we approach the same problems. More on GitHub.
Enterprise Context Layer
Turns messy company data into a clean, cited knowledge layer so AI agents answer with valid sources.
Enterprise SLM Data Cleaner
Normalizes inconsistent SAP-style master data to one standardized convention at scale, offline, with an audit trail.
LogParser-Trail
Parses security logs into typed templates fully offline and writes one audit record per line, so each template traces back to the lines that built it.
GitHub · Paper · Hugging Face
SecOps-2k dataset
2,000 synthetic sshd, sudo, UFW and auditd lines with LogHub labels in tight and loose grouping, plus a pinned Drain baseline.
GitHub · Hugging Face · Zenodo
From first email to handover.
Email the problem
A short description, a repo, or a list of problems. We reply within one business day.
30-min call
We talk through the problem, the constraints, and what done looks like.
Scope & proposal
You get a concrete scope and a fixed price. No surprises.
Build & handover
We build it, document it, and hand it over. If we agree on a scope and we don't deliver it, you don't pay.
SAP and data work in production.
Automated the PLM migration workflow at the Hero Group.
MbitAI supported the data migration for our PLM implementation: preparing data from an existing database, converting SAP bills of materials into a new recipe format, and validating the migration files. This included mapping complex data structures, building clear rules for the new PLM model, and documenting the work so the team could follow it. Careful, strong collaborator.
Have something to build?
Currently accepting new projects. Email info@mbitai.com with a short description, a repo, or a list of problems. Based in Munich, working remotely with teams in Germany, the USA, and internationally.