About
8 years building digital products in B2B and B2G. Production deployments in infrastructures ranging from Banks and Nuclear Plants to archaeological expeditions and museum collections. 4 years building AI/ML products — two in production, a dozen MVPs and data pipelines. 7 years of engineering and analytical background in B2B infrastructure, in data centers from Miami to Almaty.
Hands-on with Cursor, Lovable, Claude Code, Google Stitch, Apify, and NotebookLM. AI-coded deliverables include anti-fraud analytics, a cloud-churn forecast for client migration, a vacancy-scoring system, and a personal OS for relocation. JTBD + Impact Mapping as a working methodology.
Currently writing a PhD in Digital Humanities; teaching and mentoring.
Experience
Product Owner · 3HCloud — B2B IaaS provider, Miami, FL, USA
- Introduced AI-coding tooling (Cursor, Claude Code) to independently close economics, analytics, and legacy-backlog tasks. Freed up >50% of the working time of 5 developers and 2 OpenStack engineers, refocusing them on infrastructure development and tech-debt reduction.
- Designed and deployed a traffic-control system and anti-fraud detector — bandwidth and colocation costs −20%.
- Built analytics pipelines on proprietary billing, ticket, and OpenStack-metrics data with Claude Code — without engineering involvement — and deployed a churn-MRR forecasting and management model. Result: despite a significant tariff increase, churn rate <7%, churn-MRR <2%, +30% MRR, +55% ARPU, and NDR from 60% to 110% in six months.
Co-founder & CTO · AI and Arts R&D Lab — European University at St. Petersburg
- Launched a research lab as a product-driven startup. Result: 2 products released, 7 experimental MVPs, 9 datasets, a GPU-cluster design for inference. Partner network: Hermitage Museum, Archaeological Institute of RAS, Yandex, 36 universities, 5 museums, 500+ researchers, 250+ students, and 50 interns per year.
- Built from 0→1, with an 8-person team, a CV/ML visual-similarity search (ResNet + CLIP) for archaeological artifacts — SIMILIS. Defined the product concept, goals, and metrics; made and defended the key decision: visual-similarity search instead of a classifier. Designed the data pipeline, validation sets, quorum expert review, and data-drift control via cross-validation. Result: 50,000+ objects, sub-second time-to-first-result, Recall@20 >90%, MAE <0.04, HITL rate <5%.
- From an analysis of the Hermitage's requirements, created the concept, PRD, roadmap, key metrics, budget, and task scope for the MVP of FECIT — a CV+LLM platform for describing and attributing engravings. Saved 10,000+ hours of collection-description work, surfaced 700+ unique Hermitage prints for new exhibitions. AI identified the contours of a Leonardo painting with 86% confidence.
- Prototyped and built data-collection pipelines for museum GIS, video analytics, heatmaps, OCR, and CV analysis of X-ray and infrared paintings imagery. Adapted university policies for AI/LLM use; launched an original course on AI and prompt engineering. Result: EUSP became Russia's first fully AI-friendly humanities university.
Technical Product Manager · DataLine — B2B SaaS / IaaS & data-center provider (now "Turbo Cloud" within RTK)
- Built the unit economics and sales pipeline for the DBaaS PostgreSQL MVP. Launched HA Cluster and Disaster Recovery DBaaS tiers. Result: CPI −40%, TTFV −60%.
- Launched "Cloud Disk" — a collaboration workspace on a Nextcloud fork running on a PostgreSQL HA cluster, integrated with Cloudian S3. Managed vendors, licensing, and SPLA pricing. Introduced Pay-per-Resource billing, 152-FZ certification for the HA cluster and Disaster Recovery, and an Office 365 migration / import-substitution process. Result: 500+ new B2B clients, ARPA up ~3× (≈ $200 → $550/mo), CR from 40% to 80%, MRR doubled in year one and +40% YoY thereafter.
Early Experience
- REG.RU (Samara, 2019–2020) — Product Manager. Launched IaaS on VMware vCloud Director. $20K MRR, payback within the first 6 months.
- Security Code — Systems Analyst → TPM (virtualization security), 2017–2019. Owned requirements, user stories, and use cases for the vGate security product. Introduced UX testing and a design-to-development process, cutting release TTM from 2 years to 6 months and frontend-feature TTM from 3 months to 2 weeks.
- Altum Group Integration — Co-founder (2013–2017). Infrastructure startup; enterprise virtualization projects.
- Repin St. Petersburg Academy of Arts — Lead Engineer (2007–2013). Built the institution's IT infrastructure from scratch: network, servers, virtualization.
Key Competencies
AI/ML in production
Computer Vision (ResNet, CLIP, U²-Net, SIFT), LLMs (Qwen, Claude, GPT-class), RAG, OCR, Fine-tuning, Vector DBs, MCP / Agentic, n8n, Prompt & Context Engineering.
AI-native PM tooling
Cursor, Claude Code, Apify, NotebookLM, Lovable, multi-agent systems, Ollama, vibe coding, AI-assisted prototyping.
Product management
0→1 launches, Discovery, Go-to-Market, Impact Mapping, Advanced JTBD analysis, unit economics, pricing strategy, HADI, competitive intelligence, risk assessment, stakeholder management, B2B / B2G GTM, HITL design, sales support.
Cloud infrastructure (background)
B2B IaaS / PaaS / DBaaS, OpenStack, PostgreSQL HA, S3, HA/DR solutions, 152-FZ (Russian GDPR-equivalent) compliance, SLA design.
Education & Languages
Beyond product
Infrared photographer and historian of infrared photography — I shoot what the eye doesn't see. The same instinct that finds the unseen signal in an image is the one I bring to products.