AI and fullstack engineer who designs generative-AI platforms and ships them to production
A macro view inside a grand piano: felt hammers resting against steel strings.
AI and fullstack engineer who designs generative-AI platforms and ships them to production
Leads the architecture of AI platforms and still builds them: agentic systems, retrieval pipelines, the AWS and Azure infrastructure underneath and the interfaces on top. Delivering generative-AI solutions since 2018. Lecturer for emerging technologies and AI at a German university of applied sciences.
Role I
Employed at an infrastructure technology startup
Object detection and semantic segmentation models for automated visual inspection from image data, from model research to cloud inference.
Technologies: PyTorch, TensorFlow, Flask, CUDA, AWS, Azure
Role II
Freelance for a mid-sized chemicals company
Own AI and IT consultancy for mid-sized companies, mainly in chemicals manufacturing and its logistics: application ownership, logistics automation and applied NLP.
Technologies: Java, Swing, Python, SQL, Oracle, PL/SQL
Role III
Founded an education technology startup
Founded and built a generative-AI learning platform for IT professionals, funded by an EXIST founder scholarship: web app, native Android and iOS apps, and the ML pipelines behind them.
Technologies: AWS Fargate, Amazon RDS, Amazon CloudFront, Amazon S3, Amazon Route 53, AWS CDK, Amazon SageMaker, Apache Kafka, Git, T5, Flan-T5, LayoutLMv2
Role IV
Teaching at a large higher education institution in Germany
Teaches a course on emerging technologies and applied AI at a German university of applied sciences, on a curriculum designed from scratch.
Technologies: Machine-learning frameworks, Cloud ML platforms
Role V
Employed at a large education technology company
Built generative-AI services for an education-technology company on its existing AWS platform: fact correction, web research, information extraction, quality control and content tagging.
Technologies: Python, TypeScript, Node.js, SQL, AWS CDK, AWS Lambda, Amazon SQS, Amazon SNS, Amazon RDS, Amazon Aurora, React, Next.js, Pinecone, OpenAI API, Sentry
Role VI
Consulting for a global technology enterprise
Solution architecture for AI-assisted, large-scale data classification and labelling at a global technology company: a hybrid supervised and generative-AI system with an adaptive feedback loop.
Technologies: Python, SQL
Role VII
Consulting for a global manufacturing enterprise
Lead architect and AI lead for a global manufacturer's internal AI delivery unit, engaged through a global professional services firm. Defined the target architecture for more than ten AI use cases and delivered several end to end.
Technologies: LangGraph, MLflow, Azure AI services, Snowflake, React, OAuth, REST, WebSockets, Web scraping, Claude Code (CLI and SDK)
Every technology named in the role records, by family, with the roles whose record names it.
The full CV is available with an access code.
Research briefs, brain sparks and deep dives on AI.
Stanisław Lem's scientists build a whole discipline around a living ocean and never learn what it thinks. Interpretability research faces the same question about language models. What the evidence says, and where the comparison fails.
Four improvisations at the piano, about 14 minutes in all. Each one grows a plant in the dark while it plays.
Specimen 1 of 4
When it has played through, the plant has 765 nodes, 80 blossoms, 157 leaves and 106 shoots.
