A working log of data-engineering problems I solve to stay sharp — coding, SQL, and system design — each with notes on approach, complexity, and tradeoffs. Public on purpose: "here's the pipeline I designed last week" beats "trust me, I know this."
By George Bakaev — senior data engineer (GCP & AWS). bakaev.dev · GitHub · LinkedIn
Real-time clickstream ingestion + IP scoring @ 100k events/sec » — a full system-design write-up: requirements, back-of-envelope, architecture, deep dives (dedup, idempotent scoring), reliability, cost (~$7–9k/mo), and what changes at 10×.
coding/ one folder per problem: solution + notes
sql/ advanced SQL — window functions, CTEs, performance tuning
system-design/ design write-ups with architecture diagrams and cost numbers
templates/ notes template for new entries
- Interview conditions: timed, plain editor, no autocomplete, no lookups.
- State the approach before writing code.
- Every entry gets notes: approach, complexity, what I'd do differently, follow-ups.
- Ship the scrappy version, refine later.
| Date | Problem | Type | Result |
|---|---|---|---|
| 2026-07-06 | Clickstream ingest + IP scoring @ 100k/sec | system design | complete |
| 2026-06-05 | Top K Frequent Elements | coding | accepted, first submit |