Expertise
I donβt believe in IT directors who read about neural networks in the news. I believe in the ones who run docker compose at 5 a.m. to test a RAG pipeline β then explain to the board why this particular LLM will cut 30% of operating costs. The four pillars of my practice are below, each with numbers.
IT & architecture
Technology should outlive the hype. I design systems that live in production for years β from Hotmail voicemail to LLM orchestration on owned infrastructure.
- Expedia: led the development of DBMS management tooling β downtime across all services cut from "several times a year" to a minimum.
- Investment fund: a Gen AI / LLM / MCP pilot cut the analytics prep cycle from weeks to hours.
- RAG on owned infrastructure: data stays inside the perimeter, models under control β not in someone elseβs cloud.
How I apply this for clients
I start with an architecture audit: where the system breaks at 10Γ growth, where data leaks, where you pay for cloud you donβt need. Then a roadmap prioritized by economics, not by fashion.
Relevant service: Tech DD / architecture audit β
Teams & management
An IT function is people and processes, not an org chart. I built teams from zero three times β and I know what every hiring mistake costs.
- Three IT functions from zero: from the first laptop to 50+ people, CI/CD, and cloud infrastructure.
- Microsoft: coordinated distributed teams across Russia, India, and the US β daily sync across time zones.
- Agile/Scrum with DevOps elements β implemented as a scrum master at Microsoft and as CTO at the fund.
How I apply this for clients
I assess your current team honestly: who pulls, who drags, which roles are missing. I build hiring, onboarding, and KPI processes so the system works without an "irreplaceable team lead".
Relevant service: Fractional CTO (retainer) β
Product
Technology without a product is just cost. I build my own AI products so I see the technology through the userβs eyes β and its metrics β every day.
- An LLM-based psychotherapy support platform (Flutter + NestJS + FastAPI + RAG) β a live product with real activation and churn metrics.
- An NLP-interface assistant for tutors β time saved on routine measured in hours per week.
- "LLM-first prototyping" practice: a fast PoC on real data in days, not months, before any large-scale rollout.
How I apply this for clients
I translate "we want AI" into a product statement: which user problem it solves, how we measure the effect, what the MVP is. A prototype on your data before you could even approve a spec with a vendor.
Relevant service: AI audit β
Business strategy
IT strategy exists only when tied to money: CAPEX, OPEX, TCO, payback period. Everything else is slideware.
- Symetra: a three-year IT strategy with best/worst-case scenarios β built on TOGAF and ITIL.
- Full IT budget ownership: CAPEX/OPEX, licenses, cloud spend, payroll for teams up to 30+.
- The TCO argument for the fractional model: a full-time CTO costs β½650Kβ1.3M/mo versus a β½150β200K/mo retainer for ~20 h/mo of actual work.
How I apply this for clients
I calculate the economics before the technology: the cost of a mistake, the cost of inaction, and whether itβs cheaper to hire, buy, or build. The result is a strategy you can defend before the board with numbers.
Relevant service: Express diagnostics β
First step: a 30-minute diagnostic call
Weβll break down your task, and Iβll tell you honestly whether you need a CTO at all.
Message on Telegram