Stefan Baum
Senior Software Engineer · System Architect · DevOps
Senior software engineer and system architect, 20+ years. Full-stack development, legacy modernization, containerization, DevOps. Since 2025 as private R&D: local AI infrastructure and workflow integration.
Enterprise IT, automotive, public sector, e-commerce, travel. Freelance since 2019.
Services
Full-Stack Development
End-to-end web applications, REST/GraphQL APIs, and microservice architectures.
DevOps & Infrastructure
Build pipelines, container orchestration, and production monitoring.
System Architecture
Architecture design and review for distributed systems: integration, event-driven design, API contracts, and technology decisions.
Legacy Migration & Modernization
Assessment, containerization, and stepwise migration of legacy systems, from VM hosting into containers and through staging to the production cutover.
Technical Team Support
Hands-on support for development teams: code reviews, pairing, knowledge transfer, and continuity through migrations and staffing changes.
Embedded Systems
In-vehicle infotainment systems for automotive.
Selected Projects
Embedded Platform: Infotainment
In-car infotainment development for premium automotive brands, across multiple device generations in a large multi-vendor setup. Prototyping, feature development, and project coordination.
Enterprise Java Containerization
Migration of a legacy Java booking platform from VM hosting to a containerized Docker architecture: discovery, staging, production cutover, and stabilization afterwards. Full knowledge transfer and ongoing support for the client's technical team included.
E-Government Digitalization, Baden-Württemberg
Digital transformation platform for public-sector services in Baden-Württemberg. Frontend architecture and feature development with Angular 14 and NgRx.
E-Commerce Data Integration
B2B data aggregation for an e-commerce group: a MuleSoft integration layer, a Navision data provider, and ETL pipelines between shop and ERP systems.
Travel Booking Platform
Java reservation system for bike tourism: booking platform, admin tooling, and website, full-stack from the data model to the frontend, technical support included.
High-Volume Transaction Platform
Full-stack internet booking engine with GDS backend integration, operated 24/7 at high transaction volume. A decade of production ownership: performance optimization and high availability.
AI-Assisted Workflow
I use AI tooling as a productivity layer inside a disciplined engineering process. I review and test every result before it ships. The tools speed up the work, the engineering standards stay the same. The local stack below is private R&D, not client work.
Claude Code / OpenCode / Hermes
Agentic coding CLIs for implementation, refactoring, review, automation, and system integration.
VS Code / Cursor
AI-assisted IDEs for day-to-day development.
Local LLM Inference
I serve open-weight models with vLLM, llama.cpp, and MLX on my own GPUs: NVIDIA Ampere-class cards, a dual DGX Spark pair running tensor-parallel, and Apple Silicon for local development and data analysis. For projects and data that must stay on-premise.
Model Evaluation
I compare current open-weight models for development and document-heavy workloads: quantization (NVFP4, GGUF), context length, throughput, and how reliably each one calls tools.
LiteLLM
One OpenAI-compatible gateway in front of local and cloud models. It authenticates every request, routes sensitive prompts to local models by privacy class, and keeps fallbacks and spend limits in one config.
LangChain / LangGraph
LangGraph runs the document pipeline, LangChain for LLM integration.
RAG on PostgreSQL
Retrieval-augmented generation with pgvector. Semantic search over the project's own data.
Document Processing
A staged pipeline that takes in business documents, extracts content with OCR and vision models, classifies each one, and files it into a document management system and a RAG index.
Playwright
Browser automation for end-to-end testing and workflow automation.
MCP Integrations
Model Context Protocol servers connecting AI tooling to internal systems and data.
Context Engineering
Structured context for agent workflows: curated context loading, machine-readable session logs, and guardrails that keep agents grounded in project state.
About
Senior software engineer with over 20 years in full-stack development, system architecture, and DevOps. All-rounder: distributed systems, legacy modernization, containerization.
As private R&D since 2025: local AI infrastructure on a multi-node GPU environment for open-weight LLM inference and AI-integrated workflows.
Diplom-Informatiker (FH), Fulda University of Applied Sciences, Applied Computer Science. Languages: German (native), English.
Technologies
Contact
Available for projects as a freelance Senior Software Engineer & System Architect.
- Email: info@s-baum.dev
- Phone: +49 162 2906782
I use your data only to answer your inquiry and do not share it with third parties.