AI transformation, built on engineering discipline. Engineering leader in Munich — former VP of Engineering and CTO, 20+ years in software, 14 leading teams. I build AI-native delivery systems and the reliable AI platforms they run on, and stay hands-on enough to validate the architecture, the code and the production model.
- AI-native delivery. Specialized agents execute scoped work through GitHub Issues, which keep each task's context and decisions between sessions; architecture, priorities, risk, review and acceptance stay human-owned.
- Engineering leadership. Growing engineering organizations, product trios, hiring and delivery systems — from 5 to ~40 people at Homeapp.
- Reliable AI platforms and SRE. LLM gateways, Kubernetes, observability and SLOs, modernizing legacy stacks without stopping delivery.
| Snappy Loop | AI language-learning product, built and run solo on an AI-native delivery model in 2024–2026. 200 → 5,000 users in three months; load-tested at 20k RPS. |
| Homeapp | As VP of Engineering, scaled engineering from 5 to ~40 people with product trios, Scrum, hiring and a Kubernetes platform. |
| CHECK24 | Led a six-month backend modernization and SRE transformation — bare metal to Kubernetes — recognized with the internal Best Client-Oriented Product award. |
- Keystone (KTMS) — open-source, self-hosted LLM gateway: effort-based routing, fallbacks, spend caps, local GPU nodes. Grew out of the Snappy Loop platform.
- light-tokenmeter and ktms-admin — usage reporting and the web console around it.
- One Bot, One Pod, One Problem at a Time — SREday Munich 2026. Scaling a Telegram bot on Kubernetes: webhooks, Redis, SLOs.
- Stop Prototyping, Start Shipping — meetup at Google, December 2025. Taking AI features from prototype to production.
vasiliy.pro · LinkedIn · Email