AI agents for a data team, without handing them the keys to production
What I built in 2026 for a 12-engineer billing data team, what the agentic code reviewer actually caught, and which guardrails mattered.
Ivan Blinov
Data & AI Platform Architect · EN · ES · FR · PT · RU
Remote · Americas (UTC−3 to UTC−6)
I help teams put AI agents to work on their real data and infrastructure, safely, and make money-critical data platforms fast and correct.
- Tech lead for the billing data of millions of sellers at one of the largest marketplaces in Eurasia (100+ TB).
- Built my team's AI-agent platform: 100+ MCP tools, an AI on-call agent and an agentic code reviewer, behind capability-level guardrails.
- Co-founder and CTO of Groupsi, a health-tech startup in Brazil. Earlier, engineering manager of a team that grew from 6 to 20.
contact@ivanblinov.com · LinkedIn · GitHub · CV
For humans Case studies, notes, the story behind them, and a CV.
For agents Markdown at every URL, llms.txt, resume.json and a read-only MCP server.
Selected work
Idle CPUs, 60-second queries: a global lock in ClickHouse 23.8
How a dashboard widget that needed two numbers took a ClickHouse cluster down for 48 minutes, and why the fix was a cheaper query rather than more hardware.
665 GiB of dictionaries: when ClickHouse dictionaries stop scaling
Nine hashed dictionaries quietly grew to two thirds of a terabyte per node, and every reload tried to build a second copy. How we found that only five views used them, and replaced them with joins that read 17 granules out of 282,102.
Airflow tasks stuck in "queued" forever after a twelve-second database blip
A short network flap between two datacenters left a task in "queued" for two and a half hours, and its DAG with it. The cause is a lost executor event: SELECT … FOR UPDATE SKIP LOCKED skips exactly the row it needs. Why restarting the scheduler is the wrong fix.
Expired certificates broke production twice, and nothing turned red
Once as a load-balancer black hole that swallowed 23 of 24 connections for fifteen hours, once as a silent 30-hour replica lag behind a database that kept answering. Both times the metrics existed and the alerts did not.
Notes
- LowCardinality in ClickHouse: I measured the rule of thumb, and it was wrong both ways
Note · 2 min · "LowCardinality helps between a thousand and a hundred thousand distinct values, and hurts otherwise." A code review quoted this rule, so I measured it on ClickHouse 23.8. LowCardinality wins at any small number of values, even two, and starts losing badly at about twenty thousand per part.
- Partitioning by modulo in ClickHouse: the DDL I wrote that made queries 13.7 times slower
Note · 2 min ·
PARTITION BY supplier_id % 100sounds like it spreads the load evenly. In practice it destroyed primary-key pruning: a lookup read 64% of the table instead of 3%. Here are the measurements, and why it happens.
Open source and research
- BendKV — a Redis-compatible server in Bend2 with a formally verified core: 68 proven laws, 1,769 of Redis 7.2.16's 1,812 own tests passing. Built with a team of AI coding agents; differential testing against Redis independently found nine bugs in Redis and Valkey.
- heilbronn-triangle-bounds — conditional Heilbronn triangle bounds with Lean proofs and exact certificates.
- Open Kernel Challenge 2026 — LLM-inference kernels on an NVIDIA H200 (FP4 mixture of experts, attention, the full decode layer), built with a fleet of agents. Final standings in November 2026.