Skip to content
View grgbkv's full-sized avatar

Block or report grgbkv

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
grgbkv/README.md

Georgy Bykov

Data & business analyst in Prague. I build data products end-to-end — sourcing, verification, analysis, delivery — and I care most about numbers that survive checking. Most recently I built and operated a 12-country startup-intelligence pipeline during an internship with Microsoft's Enterprise Partner Solutions team in Central and Eastern Europe and Central Asia.

Projects

SUDigger — signal-driven startup intelligence · interactive walkthrough A production pipeline that discovered, verified and scored partnership-ready startups across 12 countries and published the daily dashboards the partner-solutions team actually worked in. It flagged six funding rounds 4–22 days before public news — receipts with public links in the repo — at roughly 5% of the cost of a naive full-refresh design. Ran on a local scheduler first, then migrated to a cloud pipeline on Azure (Container Apps jobs + AI Foundry + Azure OpenAI). Ships a runnable, zero-dependency mini-engine that replays the mechanism on synthetic data.

Financial inclusion 2011–2024 — Global Findex 2025 A balanced-panel analysis of the World Bank's newest Global Findex, validated series-by-series against the official aggregates embedded in the data, plus a working paper quantifying four aggregation pitfalls that individually bias levels by only 1–2 points — and compound into sign-flipped trends. Grew out of my bachelor's thesis; auditing my own earlier version supplied the error catalogue.

Smartwatch price analysis An end-to-end pricing analysis whose headline lesson is honesty: removing duplicate-row leakage turned an impressive-looking R² ≈ 0.8 into a defensible 0.42, and the write-up explains why the smaller number is the better result.

What I'm aiming for

Roles where data quality is a feature, not an afterthought — data analytics, analytics engineering, and applied-AI work, with a deliberate push toward cloud systems architecture and solutions. The systems I build increasingly live in the cloud: SUDigger runs on Azure (Container Apps jobs, AI Foundry, Azure OpenAI, blob-backed storage), and the architecture questions behind that migration — what belongs in a managed runtime, what stays a scheduled job, where state lives — are exactly the work I want more of. On top of that foundation I'm deepening agentic-systems engineering: harness design, verification gates, and cost-aware LLM pipelines.

📫 georgybkv@gmail.com · @grgbkv on Discord / Telegram / Instagram · Prague, Czech Republic

Popular repositories Loading

  1. grgbkv grgbkv Public

    Config files for my GitHub profile.

  2. sudigger-showcase sudigger-showcase Public

    Signal-driven startup intelligence — architecture, receipts, runnable mini-engine (12 countries, 22-day latency edge)

    Python

  3. smartwatch-price-analysis smartwatch-price-analysis Public

    What drives smartwatch prices? End-to-end analysis with an honest treatment of data leakage and encoding pitfalls (pandas, scikit-learn)

    Jupyter Notebook

  4. financial-inclusion-findex-2025 financial-inclusion-findex-2025 Public

    Financial inclusion 2011-2024 on the Global Findex 2025: balanced-panel analysis validated against official aggregates, plus a working paper quantifying four aggregation pitfalls

    Jupyter Notebook