How can I help you?

Christian Holbein

Data Engineering & AI · Berlin

I help companies move their data reliably from A to B — and use AI to get more out of it.

About

I have been running CHnet Solutions as an independent consultancy from Berlin since 2018. Before that, I worked as a freelance ETL developer for several years — so I have spent over a decade at the intersection of data, systems and business logic.

My engagement model is unusually steady: contracts typically run over several years, often as an integrated part of internal teams. I am currently in my fifth year supporting a data platform team at a German premium automotive manufacturer.

Two things matter to me on every assignment: that the solution genuinely fits the organisation, and that it still works after I leave.

What I do

Data Engineering & ETL

Pipelines that move millions of records every day — without anyone having to get up at night. The focus is on production-grade ELT/ETL solutions in cloud architectures, with emphasis on observability, cost control and clean handover to operations teams.

Typical work:

• Connecting new source systems (SAP, Oracle, MSSQL, REST, S3)
• Migrating existing Talend pipelines to modern cloud-native equivalents
• Job orchestration with AWS-native services (ECS, Lambda, EventBridge, RDS)
• Monitoring & logging strategies (CloudWatch, Apache Pulsar)
• Audit tooling, runtime analysis, slot optimisation

AI Consulting & Software Engineering

Not every problem needs a custom model. More often, the pragmatic use of existing LLMs combined with solid software engineering is the right answer. I help from concept through productive operation.

Typical topics:

• Agent-based workflows on Claude / the Anthropic API
• Custom Skills, MCP servers and tool integrations
• Automating recurring engineering tasks (job onboarding, audits, reporting)
• Code review pipelines & SonarQube setups
• Infrastructure as Code (AWS CDK / TypeScript)

The two areas increasingly overlap: many of my AI projects produce tools that plug straight into existing data pipelines — for example audit bots that compare nightly job runtimes against their cron schedules and drop an XLSX report in the morning.

Stack

Cloud & Infrastructure AWS · CDK (TypeScript) · EC2 · RDS · S3 · ECR · ECS · Lambda · EventBridge · CloudWatch · Secrets Manager · Systems Manager

Data & Pipelines Talend (Studio & Cloud) · PostgreSQL · Oracle · MSSQL · Apache Pulsar · NoCoDB · Jasper Reports · SAP RFC · REST APIs

AI & Automation Claude / Anthropic API · MCP servers · Custom Skills · Agentic workflows · Prompt engineering · LLM-driven tooling

Engineering & DevOps TypeScript · Python · Node.js · Git · GitLab CI/CD · ESLint · Prettier · Husky · SonarQube · Docker

Selected projects

Data platform · Premium automotive OEM (ongoing, since 2021)

Working on the data exchange backbone of a German premium automotive manufacturer (client name under NDA) for production, logistics and after-sales systems. Onboarding of new SAP and Oracle sources, migration away from a Talend Cloud pipeline towards container-based jobs on ECS and Lambda, and the design of an audit framework that compares runtimes with cron plans to surface bottlenecks. Current focus: AI-assisted job onboarding automation and SOPs for cloud operations. Engagement length: 5+ years, integrated into the internal platform team.

Shipcon · E-commerce logistics (ongoing)

AWS-based backend platform for a shipping-logistics provider. Split across multiple CDK stacks: carrier integration (shipping-API connectors), Shopify connector, billing layer and shared infrastructure. Combines the e-commerce order flow with carrier and tracking APIs under a single AWS architecture — Lambda, RDS, S3, API Gateway, all orchestrated via CDK in TypeScript.

Meinewelten · Premium manufactory ERP

Migration of a grown PostgreSQL application database into a consolidated ERP RDS instance, in parallel with building a new ERP schema. Accompanying data-model refactoring (legacy 1:1 copy into meineweltendb_legacy, refactored meineweltenDB for the new ERP app). NoCoDB as the internal admin UI on top of the legacy view, hosted on EC2, automated via CDK.

CHnet-Infra · Own platform

AWS CDK project that turns the entire infrastructure behind chnet.solutions into reproducible code — static Astro site on S3+CloudFront, NoCoDB on PostgreSQL RDS, and SonarQube. Includes automated cost saving via EventBridge schedules, Secrets Manager for credentials, and IMDSv2 as default. Not a showcase but a real-world setup: I use my own platform as a lab for patterns I later recommend to clients.

CHnet AI Bot · Live on chnet.solutions

Vita chatbot answering questions about services, skills and availability — built on AWS Bedrock (Claude Haiku 4.5 via the EU cross-region inference profile). Defense in depth: Bedrock Guardrail with denied topics, hardened system prompt with a refusal anchor, per-IP rate limit on a DynamoDB counter table, AWS Budget chain with an auto-disable Lambda as a cost cap. Frontend: Astro island, no tracking, EU data residency. End-to-end CDK, deployed via GitHub Actions OIDC.

Earlier engagements

Pharma · banking & finance · telecommunications · fashion and jewellery retail. Mostly ETL and reporting work, frequently spanning multiple years.

Industries

Automotive · pharma · banking & finance · telecommunications · retail (fashion & jewellery) · e-commerce logistics

How I work

Long-term. Engagements last years, not weeks. That is intentional — complex data landscapes don’t reveal themselves in a single sprint.

Remote-first. Based in Berlin, mostly remote work, on-site visits within DACH where needed.

Direct. I work without an intermediary layer. The person you talk to is the person who writes the code.

Handover-ready. Whatever I build should keep running without me. Documentation, IaC and runbooks come with the package.

Certification

AWS Cloud Quest: Solutions Architect — Credly Badge