Hi, I’m Nils, an AI-native engineering leader with 6+ years leading engineering teams and 15+ years building at the intersection of AI, data and complex systems, most of it remote. I founded and led the first AI team at Pitch, and at Phaidra grew from leading a single AI team to running a 30+ person engineering org, building teams from scratch and evolving the organization as it scaled. And I still ship agentic systems myself, every day.

I move comfortably between people management, technical leadership, and hands-on engineering, and I’m at my best when I can flow across all three. In 2026 I leaned into the building: two production LLM systems launched, one as a founder-CTO and one from working prototype to production in six weeks, with deep work on agentic developer workflows, evaluation and the data flywheels around them. My working belief: code is cheap now, and verification and reliability are the real constraint. I run engineering on that model, humans directing agents with verification built into the loop.

What drives me is pointing software, AI and data at problems that matter: health, sustainability, and the systems our societies and environment depend on. I’ve worked on everything from reinforcement learning that cuts industrial energy use to data platforms that support sustainable investment.

Collaboration and continuous learning are at the core of how I work. I believe the best solutions come from diverse teams and ideas, and I try to build environments where people do their best work. Thanks for stopping by. If any of this resonates, I’d love to connect.

Contact

I’d love to hear from you. Whether you’re a startup, an agency or an enterprise: if you have an interesting engineering-leadership role, or are looking for hands-on help with AI and agentic systems, please get in touch.

I speak German (native) and English (fluent), plus basic Swedish and Italian.

eMail: nils@blum-oeste.de ( PGP key) Phone: +49 151 152 282 62 LinkedIn

Expertise

Engineering Leadership

  • Leading engineering at Director / Head-of scope; managing managers
  • Org design and building teams from scratch; company-wide career ladders and performance frameworks
  • Hiring at scale; coaching ICs into Engineering Manager and Tech Lead roles
  • Remote-first organizations distributed across India, Europe and North America

AI-Native Engineering Practice

  • Agentic engineering and the agentic SDLC (daily production practice)
  • Evals and evaluation-driven development
  • Agent orchestration; human-in-the-loop oversight design; guardrails and safe automation
  • LLM observability and tracing (Langfuse, OpenTelemetry); context engineering

AI/ML and Software Systems

  • Agentic LLM pipelines and document intelligence
  • Reinforcement learning, from research to production
  • Distributed, real-time and data-intensive systems
  • Software architecture; reliability and release engineering

Programming and Platform

  • Python, Clojure/ClojureScript, TypeScript, SQL
  • Agentic SDLC tooling (Claude Code, MCP, custom harnesses); LLM APIs (Anthropic/Claude, Vertex AI)
  • Kubernetes, Terraform, Docker, CI/CD; GCP, AWS
  • PostgreSQL, Kafka, Snowflake; Grafana, Prometheus; Linux, Git

Data and Machine Learning

  • Data engineering and data flywheels
  • Relational and non-relational databases
  • PyTorch, MLflow, DVC; classical ML and deep learning
  • Scientific data analysis and statistics

Education and Certificates

  • Dr. rer. nat. (PhD) in Geoscience
  • Artificial Intelligence Nanodegree
  • Certified Scrum Product Owner
  • Certified Scrum Master

Talks

Depression - Why we need to talk more about it: A lightning talk at the Heart of Clojure conference 2019 about Depression, its impact and the stigmatisation that prevents addressing the issue appropriately and empathetically.

Intro to Probabilistic Programming and Clojure’s Anglican: My talk from EuroClojure 2017 about probabilistic programming in Clojure using the Anglican library.

Abstract: Probabilistic Programming Systems aim to merge general purpose programming with probabilistic modelling. They provide powerful statistical inference and thus allow developers to focus on the modelling with tools and environments they are comfortable with. These emerging methods are promising additions to the Data Scientist’s toolbox and an interesting, satisfying playground for programming enthusiasts. This talk is an introduction to Probabilistic Programming Systems, their use and value for the industry and Clojure’s great library “Anglican”.

Building declarative GUIs with Clojure, JavaFX and fn-fx: A gave this talk at the Clojure Berlin Meetup as an introduction to using JavaFX for building desktop applications with Clojure as an alternative to the more established way of using GitHub Electron and Clojurescript.

Microservices, Music and More: “Christoph and Nils give an overview of the Microservice architecture, as well as its benefits and potential drawbacks. They show how components like Docker play an important, enabling role for this architectural style and how the complexity of the service landscape can be managed.”

Mundane Clojure for Mortals: A Simple Web API: “Our resident LISP fan in a very hands-on tutorial about getting a web API up and running using Clojure.”

Generative Design: “Nils is getting artistic with the help of dots, lines, circles and a generous helping of Perlin noise.”