I’m Serhii, a senior engineer with 15+ years in frontend, several as a lead. I turn a vague request into a scope engineers can review, then build it across UI and API. Your team estimates real work instead of guessing from mockups.
I own the work through testing, release and production debugging, with a multi-agent workflow doing the legwork.
Open to Senior / Staff Product Engineer roles
Remote · Ukraine, EET (UTC+2/+3), overlap with EU and US East
Working principles
Research before scope
I study the domain, the legacy behavior and the code before I propose scope.
Small steps that ship
I cut work into the smallest end-to-end flows that are testable and useful on their own, and release them one at a time. No big-bang launches.
Check the whole flow
I trace the complete flow myself and confirm what I find with whoever owns the code. I don’t take a colleague’s assurance or an agent’s report on trust.
Case studies
Evaluation against customer test data
A US regional bank needed to score models against its own test data, and DataRobot offered that only in a legacy UI. I owned my team’s part of moving the supervised workflow into the new product. It shipped in about five months, and customers used it in production.
Customer-defined evaluation metrics
Customers could define their own evaluation metrics only through the API. I led the three-phase product integration: a UI for packages of up to 50 metrics, validation of each file as it’s uploaded, and the Python API changes behind both. All three phases were built behind a feature flag; one UI fix was still in review when I left DataRobot.
Multi-agent engineering workflow
I built it for my own work at DataRobot. Agents research, implement and test in isolated worktrees, up to four tracks at once. I set the scope, review every diff and accept the result.
Contact
Hiring for a product engineering role? Write to serhii@babich.dev — or grab the CV.