Currently building Swobu — a local boundary layer that decouples AI clients from any backend.
I work across the full depth of the AI stack — from ML modeling, training and MLOps, through large-scale distributed systems and reliability engineering, down to interface and frontend architecture.
This range comes from 15+ years of building complex systems across multiple layers:
- Meta — leading ML initiatives involving multimodal models, LLM agents and ranking systems.
- Mercedes-Benz & Vay — leading ML teams for robotaxi fleets, including spatio-temporal forecasting and rare event prediction.
- AWS — designing distributed systems and high-reliability data pipelines.
- Yandex & HERE Maps — building high-performance web mapping platforms and large-scale services that served millions of MAU.
This broad, hands-on experience lets me move fluidly between model behavior, distributed system trade-offs, frontend constraints, and operational realities — often in the same conversation. It becomes particularly valuable when building complex, cross-functional AI platforms where teams are usually siloed.
Selective about new opportunities. Interested in high-impact contract work, advisory roles, or founding situations in AI infrastructure, agentic systems, and complex production ML platforms.
- Email: contact[at]dmytrii.com
- LinkedIn: linkedin.com/in/dmytrii
- Twitter: @mlpreview
Happy to talk if the problem is genuinely hard and interesting.