01
Chief Product Officer
September 2024 — September 2025
HyperSpace
Turning around an AI marketplace whose roadmap could not reach users.
- Constraint
- Weak retention and unpredictable delivery
- Outcome
- Retention curve AUC 13% → 26%
Product leader
Konstantin Khalyushev
17+ years in technology · 11+ years in product
I find the constraint holding a product back, and build the system that removes it.
I have led search and recommendation products inside a 45M+ MAU media ecosystem, rebuilt monetization and international growth for a B2C mobile portfolio, and turned around an AI marketplace. Today I work close to the product: prototyping, building early versions myself, and validating with real users before committing an engineering cycle.
01
Evidence
Every number below belongs to the company it sits under.
Three constraints, three systems built to remove them.
HyperSpace
AI marketplace · product turnaround · CPO, 2024–present
Retention was weak and delivery was unpredictable, so no roadmap could be trusted to reach users.
- retention curve AUC
- 13% → 26%
Ray / ReactivePhone
B2C mobile EdTech portfolio · PM to CPO, 2019–2022
Growth was capped by a mobile-only self-image and a pricing and positioning model built for a single market.
- downloads
- 18M+
- conversion, Poland
- ~12×
- ARPU, year over year
- ~5×
- company revenue
- ~2×
Rambler&Co
Search & recommendations in a 45M+ MAU media ecosystem · 2013–2017
Search monetization and recommendation quality were limiting revenue and engagement across the ecosystem.
- search revenue, YoY
- +95%
- recommendation CTR
- 2.5×
- page depth
- 3×
- SERP conversion
- +30%
02
Ways to work together
Two ways in. Same person, different level of involvement.
01
Join a team
I have held CPO and founder titles. I am deliberately looking for work closer to the product, the users and the execution — not further from it.
- Senior and Lead Product roles
- Senior hands-on IC product work
- AI, growth and B2C product teams
02
Work with me
For teams that need senior product judgment without adding a full-time executive, or that want an outside read on where the real constraint is.
- Fractional product leadership
- Part-time senior product work
- Advisory and selected consulting
Second opinion
Would I actually be useful to you?
Do not take my word for it. Give my site to the AI that already knows your context, and ask it.
ChatGPT (opens in a new tab)Claude (opens in a new tab)Grokpaste (opens in a new tab)Geminipaste (opens in a new tab)DeepSeekpaste (opens in a new tab)
Grok, Gemini and DeepSeek can't take a prompt from a link. Those three copy it to your clipboard and open a new chat — paste it there.
Show the promptHide the prompt
I am considering talking to Konstantin Khalyushev about a product problem. Based on what you know about my work, company, product, team and current priorities, and the information available on his website, assess whether talking to him would actually be useful. Please: 1. identify which of my current problems he is likely to be useful for; 2. explain where his experience appears particularly relevant; 3. identify where he is probably not the right person; 4. suggest what I should ask him if we talk; 5. assess whether a one-off consultation, a longer advisory engagement, or neither seems appropriate. Be critical. Do not recommend him just because I asked. If there is not enough evidence that his experience is relevant to my situation, say so. His website: https://khalyushev.com/en
Your conversation stays between you and your assistant. I never see it.
How I work now
I build the first version myself.
Over the last few years my work has moved closer to implementation. The reason is speed of learning: a working prototype answers questions that a document cannot, and it answers them before a team commits a full engineering cycle.
- 01Rapid prototyping
- Building an early working version to test the idea against real behaviour rather than opinion.
- 02AI coding agents
- Using coding agents to move from a product hypothesis to something runnable in hours.
- 03LLM APIs and RAG
- Designing and wiring retrieval and generation directly, so the product constraints are understood first-hand.
- 04Evaluations
- Building evals so AI quality is measured against real tasks instead of demo impressions.
This is where my founder projects come from — they are the practice, not a side career.
03
Projects
Three cases with company-scale outcomes, plus the products I have built myself.
Selected professional work.
02
Product Manager → Chief Product Officer
March 2019 — June 2022
Ray / ReactivePhone
Breaking a mobile-only company out of a single-market growth ceiling.
- Constraint
- A mobile-only self-image and single-market pricing
- Outcome
- ~12× conversion in Poland · ARPU ~5× YoY
03
Product Manager · Search and Recommendation leadership
December 2013 — September 2017
Rambler&Co — Search and Recommendations
Search monetization and recommendation quality inside a 45M+ MAU media ecosystem.
- Constraint
- Search monetization and recommendation quality
- Outcome
- +95% search revenue YoY · 2.5× recommendation CTR
Personal and founder projects
An AI assistant over marketplace seller business data.
What exists today
- In active development as a founder-led product.
- Product scope is defined around specific seller jobs rather than general analytics.
- AI-powered web and mobile applications are built and running.
- Automated marketplace data collection and interpretation workflows are in place.
- First pilot customer signed and onboarded.
02
Tomus
Founder and Chief Product Officer
June 2023 — September 2024
- EdTech
- iOS
- Music learning
- Gamification
- Monetization
An iOS music-learning product built from zero and shipped in about a month.
MVP shipped in ~1 month
Local-first browser tools for Argentine small businesses.
What exists today
- A working set of browser-based tools, built hands-on with AI-assisted development.
- Local-in-browser processing where the workflow involves sensitive data.
- No usage figures to report — this is a small, self-built project.
Get in touch
Send a message and it reaches me directly. I read everything that is specific.