I invest real money in DeFi
and publish the risk behind it.
I'm a DeFi yield and risk analyst at Doubletop. I manage allocated capital on chain, spot in my own wallets and perp margin on Pacifica, and every position has a published risk report behind it. The rest of my week is protocol research, on-chain analysis, yield modelling and building the data pipelines that make all of it possible, plus two products I ship and run myself.
Allocated capital, run in the open as a record.
No fund, no custodian, no screenshots you have to take on trust. This is real money under management, sized so it can be published in full: a record of what I did, not a position to copy.
The $10,000 is a reference size, not a ceiling. Venues and instruments were picked for depth, so the same two pairs hold their shape at a million without changing anything but the numbers. Small is what makes it publishable. It is not what makes it work.
The rule is boring on purpose: stay neutral, stay liquid, survive. Each pair holds a short leg against a long that earns yield, so where the market goes is not the bet. The yield is.
Nothing is locked. The whole book closes in about five minutes. Most yield strategies skip that part, and it matters: if you cannot get out, the yield is not real.
It has been through a venue move, a stablecoin depeg, two protocol incidents and the crash on 10 October 2025 that dragged the whole market down. No forced liquidation, no auto-deleveraging, no loss. I sat and watched the position and the API while it happened, which is the only reason I can say that with a straight face.
- Spot in my own wallets, my keys. The shorts' margin sits on Pacifica, a decentralized perp exchange. No pooled fund.
- Sized to survive, not to win. Distance to liquidation is an alert, not a mental note.
- Written up, and dated. Every venue holding the book has a published risk report with its date, re-checked every month. They were not all written first: the research desk opened in June 2026, after the portfolio, so Pendle was written up on 30 Jun, Pacifica on 3 Aug, two days after the shorts moved there, Kinetiq on 5 Aug and Jito on 30 Sep.
Two products in production. Both mine, end to end.
Design, backend, frontend, servers and the day-to-day running of them. Not prototypes and not weekend demos: Taligate has paying users, Vitrine is open to anyone, and both have uptime to keep.
Taligate
Polymarket wallet tracking through Telegram. Follow any trader, get a message about thirty seconds after they open, close or resize a position, with entry, value and PnL. Filters by size and value, leaderboards, a morning digest, and an alert when several wallets you follow pile into the same market. Free tier plus three paid plans, billed in USDT or USDC. Public data only, no wallet connection, no keys.
Vitrine
DeFi research, rates and picks in one place, and the desk everything on this page runs on. Its public side publishes dated risk reports, live boards for stablecoins, yields, funding and lending, the stablecoin dependency map, monthly picks, a where-to-put-it finder, a daily DeFi newsletter and the live portfolio, which is one module inside it. Scanners write to Postgres every five minutes and alerts go out on Telegram. One machine, behind a tunnel, no open ports.
What the work has actually produced.
About a thousand protocol reviews, a grant program, a community built from nothing, and a research method that has held up in public. Along the way the two calls that mattered, on the rsETH exploit and the Aave freeze, landed early enough to keep capital clear.
What I actually use.
Listed because you can check it, not because it is long. Everything here shows up somewhere in the two products above.
- Protocol due diligence and scoring
- Deposit, exit and liquidity risk
- Delta-neutral and funding strategy
- Stablecoin peg and depeg analysis
- Incident post-mortems
- Grant and proposal review
- Python, asyncio, Pydantic
- Postgres, schema migrations
- FastAPI, Streamlit
- Next.js, React, TypeScript
- Telegram bots and alerting
- mypy strict, ruff, pytest gates
- Dune, Arkham, on-chain forensics
- DeFiLlama, Pendle, perp DEX APIs
- Hyperliquid, Binance, Bybit feeds
- Solana RPC, EVM tooling
- LLM agents for research pipelines
- English (full professional), Ukrainian (native)
I built this skill set on purpose.
I came into crypto in 2021 with an engineering degree and no finance training, and I have been closing that gap on purpose ever since. Nobody handed me any of this work.
The first years were community and grants. Running a 50,000-member Discord shows you how protocols fail in public. Reviewing 130+ grant applications at The Graph, then leading the review team, teaches you to tell a real plan from a good pitch. That turns out to be most of what risk analysis is.
Research came next, and the honest problem was that reading protocols by hand does not scale. So I learned to build. Python, then Postgres, then a full Next.js frontend, not to become an engineer but because the analysis I wanted was impossible without the pipeline under it. The monitoring system exists because I got tired of missing things.
Taligate came from the same instinct pointed at a different market. Polymarket wallet data nobody was serving well, so I shipped a product and started charging for it.
On paper it is a STEM master's with honours. In practice it is running my own money in the open since May 2025 and publishing the reasoning where anyone can check it.
M.Sc., with honours · Electronic Communications and Radio Engineering, Vinnytsia National Technical University. Based in Ukraine, remote since 2020.