Six years in startups as a Python backend, data and AI engineer. Today I lead the backend team of an AI-powered analytics platform, and on the side I design and operate my own products, from a self-hosted email service to a multi-tenant CRM.
Auf DeutschLead Software Engineer mit sechs Jahren Erfahrung in Python-Backend-Systemen, Datenpipelines und LLM-Integration. Ich wohne in Dortmund, besitze eine Chancenkarte und bin ab sofort deutschlandweit verfügbar. Deutsch B1, Englisch C1.
Five systems, each with what it does, what I personally built and the decisions behind it. The full write-up with architecture diagrams is in the portfolio PDF.
Multi-source marketing data for hotel portfolios, turned into reports people act on. I built the analytics and reporting modules, the security-gated CI/CD to Kubernetes, and now lead the team and the architecture.
A tool-calling LLM agent that takes implementation-ready tickets, implements them with tests on a feature branch, opens the pull request and reports the cost of every run.
A SendGrid-class platform that owns its whole mail pipeline: SMTP relay and REST sending, DKIM, direct delivery to recipient MX servers, bounces, tracking, hosted mailboxes and metered Stripe billing, on one VPS.
White-labelled lead management on one shared PostgreSQL database, with tenant isolation enforced in the ORM and again by Postgres row-level security, and tests that run as a non-superuser so the policies are really exercised.
A time-boxed evaluation of dbt for the silver-to-gold layer of the ETL v2 platform, run on our data shapes, our Postgres and our Celery orchestration, with every number reproducible from a runbook.
Independent metered API tools behind a Traefik gateway with a shared auth, metering and billing core.
FastAPI, Traefik, Postgres, StripeWordPress sites rebuilt on FastAPI and Postgres with a reproducible MySQL-to-Postgres pipeline, legacy password upgrade on first login, Stripe checkout and certificates with QR verification.
FastAPI, Postgres 16, Next.jsUpload, OCR, field extraction, a trained classifier and a LangChain agent that explains the decision.
FastAPI, scikit-learn, LangChain, Azure OpenAITools generated for every endpoint in the platform's OpenAPI spec, so AI assistants can query the product. Team-built under my lead.
Node.js, MCPMulti-tenant campaigns and contacts with open and click tracking and bounce webhooks; the predecessor that led to InkPigeon.
Django REST, CeleryETL pipelines and a data warehouse, dashboards, scrapers, a recommendation system and RASA assistants fed by Google, Meta and Microsoft APIs.
Python, Django, RASA, StreamlitSix years in startups: joined Cogwheel Analytics as a senior engineer when the backend was an early product, became its most active contributor and moved into the lead role as the platform and team grew.
Only what the work above demonstrates. No percentages.
Python services with FastAPI and SQLAlchemy 2.0 on PostgreSQL; Celery for scheduled and background work; ingestion from vendor APIs and files; reporting engines; bronze, silver and gold modelling; dbt where it earns its place.
Tool-calling agents with bounded loops and cost accounting; retrieval grounded in a system's own APIs; versioned prompts; LLM output rendered through existing, auditable pipelines.
Docker everywhere; Kubernetes on AWS EKS with ECR; GitHub Actions pipelines with Snyk and Trivy gates and human approval; Caddy and Nginx TLS; Prometheus and Grafana; backups, runbooks and adversarial review of authorisation changes.
Code review and release management across sprints; written architecture direction with explicit dispositions and estimates; mentoring; aligning the backlog to design documents; measuring before deciding.
Before software I studied Electrical and Electronic Engineering, with a second major in Computer Science. Four IEEE conference papers between 2019 and 2021, at RAICON, ICCIT and ICREST, on face recognition, robotics, visible-light communication and MOSFET modelling. Founding Chairman of the IEEE Robotics and Automation Society student chapter and Director of Robotics at SUST, 2018 to 2020.
Recommendations from people I have worked with, published with their name and a link to their profile after I have read them.
This section is new and still empty. If we have worked together, I would value a few honest sentences about what it was like.
Notes from the work: decisions, trade-offs and what I'd do differently. All posts
Designing a certification platform's provider API so retried batches never duplicate certificates and partners can integrate without talking to us.
The most useful thing to come out of a dbt spike was fifteen lines of CSV: one place where every vendor's channel name maps to ours.
A time-boxed spike to decide whether dbt fits our silver-to-gold layer: it did everything asked, and showed how easily it produces silently wrong data.
RLS tests that run as the postgres superuser prove nothing. How the Intelect test suite exercises the policies for real, and what that costs.
I'm looking for a full-time role in Germany or elsewhere in the EU, on-site, hybrid or remote, starting immediately. If you're hiring for one of these, the fastest way to reach me is email.
Worked with me? Write a recommendation.