Fractional CTO / CIO — retainer
Technology strategy, development, infrastructure, hiring and team growth — on a part-time basis. Covers 80% of a full-time CTO’s tasks without the cost of hiring one.
Alexander Kosykh · Fractional CTO / CIO · Enterprise Architect
25+ years designing systems businesses trust with their money: from banking software in 1990s Novosibirsk to solutions for large international companies — Expedia, Coca-Cola, Verizon, GE Capital and others. Built the IT function from scratch three times, ran corporate IT departments and delivered the full lifecycle of multi-year, business-critical IT projects. AI has been a working tool since 2017, not news from a feed.
25+ years in IT — from banking software developer in 1990s Novosibirsk to CTO with an international background: Microsoft, Expedia, Accenture, Symetra. Full cycle completed: wrote code for distributed systems, designed Big Data platforms, built the IT function from scratch three times, ran corporate IT departments and delivered the full lifecycle of multi-year, business-critical IT projects.
For me, AI is not news from a feed but a working tool since 2017: bot agents and NLP pipelines at Accenture, ML experiments in insurance, production LLMs today. I work hands-on with a modern stack — LLM, RAG, MCP, self-hosted models (Qwen/DeepSeek, YandexGPT, GigaChat) — and I know how to deploy it safely: 152-FZ, data within the perimeter, integration with 1C and Bitrix24.
Particular expertise — medicine and biotech: three years as CTO of a genetics company, НЦГИ, and the very start of my career at the Institute of Cytology and Genetics, SB RAS. Today I combine two CTO roles in parallel and build my own AI products — so that every day I personally face the limitations of technologies, not just set the direction.
I work in a fractional format — and open to full-time engagements where IT is a source of competitive advantage, not just "business support".
Technology strategy, development, infrastructure, hiring and team growth — on a part-time basis. Covers 80% of a full-time CTO’s tasks without the cost of hiring one.
Fast assessment of IT and AI maturity: three main risks, a roadmap with priorities and clear economics. An honest answer on whether you need a CTO at all.
Scenario selection, RAG pipelines, AI agents, self-hosted infrastructure. From prototype to production — with KPIs fixed before the start.
Technical expertise for deals and investors: architecture, code, team, tech debt. Bringing stalled projects and teams back to stable operation.
Building an IT function from scratch: processes, stack, team. Proven three times — in data centers, biotech and AI products.
Import substitution, 152-FZ, private LLM infrastructure: your data never leaves your perimeter. Architecture that passes the security review.
A full-time CTO in Moscow costs 650 k — 1.3 mln ₽ per month — plus 3–6 months of search, contributions and the risk of a bad hire. A fractional CTO retainer covers 80% of the same tasks for 150–360 k ₽. The 3–5x difference is the main argument for the format.
The market pays not for technologies but for reducing the risk of failure — and that is exactly what a senior architect sells.
| Parameter | Full-time CTO | Fractional CTO |
|---|---|---|
| Monthly cost | 650 k — 1 mln ₽ (up to 1.3 mln with taxes and bonuses) | 150–360 k ₽ in 20–40 h/mo packages |
| Contributions and bonuses | +30% insurance contributions after SMB benefits were cancelled | None — contract with a self-employed person / sole proprietor |
| Time to result | 3–6 months of search and onboarding | Start in 1–2 weeks, monthly contract |
| Risk of a mistake | High: replacing a C-level takes months and costs reputation | Exit with 30 days’ notice |
| Annual savings | — | ~5–10 mln ₽/year versus the full-time rate |
Market price benchmarks, Russia, 2026. Audit cost is often credited toward subsequent implementation.
New Hosting Company (specialized data centers), НЦГИ (biotech), Likematica.ai (AI products) — processes, stack, team, infrastructure.
From bot agents and NLP pipelines at Accenture to LLM/RAG/MCP today — continuous practice, not retraining on the hype.
Enterprise-level projects under full management; global teams of 20–25 people on three continents.
A product taken from early drafts to a production release: a team of 50+ engineers, coordination across the US, Russia and India.
UpStudent — an AI assistant for tutors (NLP); SamBoss — a platform for delivering individual services.
Certified Enterprise Architect: architecture and processes are not separate consultants but a single decision-making system.
"AI for work, not for demo"
principle of the practice
We automate processes that actually eat up the client’s time and money — rather than stretching ready-made tools onto someone else’s processes. Against hype, for measurable results: KPIs are fixed before the pilot starts.
Clients: Verizon, GE Finance, Teradyne. Architect of a real-time geospatial data warehouse for a major telecom — 2 years, a team of 25 people.
All-Russian IT projects; a sales data collection system from dozens of sites with management reporting generation.
Banking software for remote client operations; my main build was the remote client for accessing banking systems.
Algorithms for analyzing medical and biological data: cluster and regression analysis, relationship graphs — first hands-on data science experience.
Faculty of Mechanics and Mathematics · applied mathematics and statistics · 1993 · Novosibirsk