Description

I build and ship end-to-end AI products, then hand you infrastructure that keeps working after I leave. My flagship is Vault: Privacy Guardian, a privacy-first iOS app with paying subscribers that runs sensitive-content detection entirely on-device. Alongside it: NDS Ghost Hunter (a consumer ML app, also paying subscribers), a production multi-agent LLM platform self-hosted with full IaC, and Postgres-at-scale plus knowledge-graph systems, delivered with an AI-augmented workflow that ships what normally needs a team. I work solo by default and bring in a small, vetted team for larger builds, so I can take on scope beyond a typical freelancer without diluting quality.

Before founding Westover Labs I spent a decade in US big tech: staff engineer at Uber (navigation at 10M+ requests/sec, founded the Python Platform team, led the company-wide Python 2 to 3 migration) and engineering manager at HERE building autonomous-driving capture systems from 2 to 14 engineers.

Physicist who writes production systems. MBA who speaks business. I own the whole problem across AI infrastructure, on-device ML, iOS, and backend, and I ship it. EUR 185/hr equivalent, billed as a day rate. Berlin, English native, German B2.

Languages

English
Native or bilingual
German
Basic

Workplace preferences

Can work on-site

Berlin (up to 50km)

Westover Labs UG
Founder & CTO
January 2021 - Today (5 years and 5 months)
Berlin, Germany
Shipped Vault: Privacy Guardian (iOS), the flagship product: privacy-first sensitive-content detection
• • running 100% on-device (ONNX / CoreML), with paying subscribers, live on the App Store. Images are analyzed in memory and never leave the device. Shipped NDS Ghost Hunter (iOS), a consumer ML subscription app with paying subscribers, running real on-device ONNX inference with no cloud ML API in the hot path. Built a self-hosted, multi-agent LLM orchestration platform with full infrastructure as code (Ansible, Cloudflare Tunnels, Tailscale, self-hosted CI runners): intent-routed task delegation, automated pull requests, Jira workflows, and a federated knowledge graph running around the clock. This is the deliverable I customize for clients. Consult as fractional CTO and architect on fixed-scope engagements. Stack: Python / FastAPI, on-device ML (ONNX Runtime), RAG plus pgvector knowledge graphs, Ansible IaC, React Native / Expo.
Python FastAPI Machine learning artificial intelligence PostgreSQL
Navigation, Uber
Staff Software Engineer
February 2021 - October 2022 (1 year and 8 months)
Remote Remote, Mahesana, GJ, India
Worked on Navigation, one of Uber's largest services (1400+ nodes per data center), building on-device
• • routing-ingestion software running on millions of driver phones. Raised off-route simulation capacity from 500 to 10,000 trips per run, and implemented a J2ObjC transpilation pipeline to deploy the Java detection engine to iOS without a rewrite. Led the Postmates-acquisition data-deletion strategy across multi-cloud databases with GDPR-compliant identity flows.
Location Platform, Uber
Software Engineering Manager
February 2018 - January 2019 (11 months)
Founded and built Uber's Location Platform team from scratch. Scaled Cassandra to 80+ nodes per data center at 2-10M QPS and raised GPS ingestion success from ~80% to 97%+. Led GPS anti-fraud engineering; spoofing detection prevented $30M+ in fraud. Ran GDPR compliance for Location Platform (Project Highlander single-source-of-truth consolidation).