Python automation & research tooling · Web3 data
I turn manual and repetitive processes into tested, documented, reproducible tools. My current domain focus is crypto market data and Web3 infrastructure, but the engineering work is broader: APIs, data pipelines, AI-assisted workflows, dashboards, alerts, and operational tooling. Hands-on across multiple chains — Solana, Polygon (Polymarket CLOB), Base, and MegaETH testnets — plus centralized-exchange data (MEXC, Hyperliquid, Binance).
Based in Russia. Open to 100% remote full-time or part-time roles in Python automation, AI integrations, data tooling, and Web3 analytics.
Resume · LinkedIn · Telegram · Email
- Python automation with explicit inputs, outputs, validation, and failure modes
- REST/API integrations, data normalization, SQLite storage, and Telegram-ready alerts
- FastAPI and n8n workflows with audit logs and human-review checkpoints
- operator dashboards, health monitoring, and incident runbooks
- reproducible research with tests, CI, synthetic fixtures, and documented limitations
| Project | Engineering evidence |
|---|---|
| Exchange Monitoring Lab | Paper-only Python package for normalizing MEXC and Hyperliquid data, evaluating market regime, and explaining deterministic exits. Typed models, Decimal arithmetic, synthetic fixtures, tests, and CI. |
| AI Document Review Pipeline | FastAPI webhook, explicit business rules, SQLite review queue, immutable audit trail, n8n workflow, and unit tests. Ambiguous cases are routed to a human instead of being silently accepted. |
| BotOps Control Center | Deployed TypeScript/React operations dashboard with typed health endpoints, runbooks, CI, demo telemetry, and an optional server-side OpenAI Responses API integration. |
| Solana Memecoin Analyzer | Explainable scoring pipeline with validated models, dust filtering, wallet weighting, risk gates, deterministic tests, and synthetic data only. |
| Polymarket BTC Research | Resolver QA, shadow logging, filter attribution, and analysis of 1,485 resolved observations across 181 markets. The measured hypothesis was negative and documented as not deployable. |
| AI Creator Scout | Auditable Python prototype for cleaning a creator list, transparent scoring, risk flags, and mandatory human review before any outreach. No automatic messaging. |
I use Codex and other AI tools as engineering accelerators. The workflow remains specification-led:
- define the task, boundaries, and acceptance criteria;
- build a small working slice;
- inspect logic and edge cases;
- run deterministic tests on synthetic or paper data;
- document limitations and what the evidence does not prove.
For me, “vibe coding” means fast AI-assisted iteration with human verification—not unreviewed generated code.
A working monitoring system is not proof of a profitable strategy. My public market projects are sanitized, paper-only research artifacts. They do not place orders, use live capital, expose credentials, or claim validated profitability.
Dated cohort details live in the individual repositories rather than in this profile, so reviewers can see the methodology and the historical context together.
Python · FastAPI · REST APIs · JSON/JSON-RPC · SQLite · n8n · Telegram Bot API · TypeScript · React · OpenAI Responses API · Git/GitHub · GitHub Actions · pytest/unittest
- Higher education — Saratov State Agrarian University, Engineer in Land Cadastre.
- English — technical reading and documentation with translation tools; actively improving spoken and written English.
- Interested roles — Python automation, AI integrations, data tooling, Web3 analytics, and rapid product prototyping.
- Email: gulaevoleg191@gmail.com
- Telegram: @Olejo29
- LinkedIn: Oleg Gulyaev
Создаю на Python и с помощью AI-инструментов автоматизации, API-интеграции, системы мониторинга, исследовательские пайплайны и интерфейсы для операторов. Делаю упор на проверяемую логику, тесты, документацию и честные границы результата. Ищу полностью удалённую работу в Python, AI-автоматизации, Web3-аналитике и быстром прототипировании.