Andrey Popov
I’m Andrey Popov, an AI Product Engineer focused on building reliable agentic workflows—linking LLM evaluations, guardrails, and test automation so AI features can be shipped with confidence.
About
I’m Andrey Popov, an AI Product Engineer focused on building reliable agentic workflows—linking LLM evaluations, guardrails, and test automation so AI features can be shipped with confidence. I enjoy turning ambiguous problems into safe, measurable systems: defining what should be automated, designing how results are verified, and then directing implementation with AI coding agents under clear constraints.
At EPAM Systems, I work across end-to-end AI solution engineering, from internal production ranking and candidate workflows to an AI coaching platform for leadership development clients. My background also includes finance and analytics, which sharpened my taste for provable logic and validation. I now combine that rigor with TypeScript and Python to deliver cost-controlled, evidence-based systems—often using LLM routing, tenant isolation testing, and “JSON receipt with evidence” approaches so every run is auditable.
- 21 years' experience
- Joined November 2026
Skills· 26
Languages
- RussianFluent
- EnglishFluent
- LatvianBeginner
- FrenchBeginner
Industry experience
- Software & Internet
- Financial Services
- Energy & Utilities
- Healthcare
- Manufacturing
- Professional Services
- Education
- Computers & Electronics
Work experience· 9 roles
Built and operationalized agentic LLM features with evals, guardrails, and tenant isolation testing. Owned system behavior across service-to-service and front-end surfaces, covering matching, booking, cancellation, and video-call screening via Azure. Designed validation strategies including parity gates, out-of-sample/baseline comparisons, cost-aware routing, and go/no-go release plans. Supported production ranking pipelines with large-scale model evaluation and refactoring while preserving outputs via row-level parity checks. Implemented and governed multi-model AI workflows that produce evidence-bearing JSON receipts and validated cost/quality tradeoffs.
Led statutory closing and Baltic reporting (Latvia, Lithuania, Estonia), operating under SOX 302/404 controls. Managed SAP FI/CO controlling and supported financial reporting and governance requirements.
Performed FP&A, product costing, and margin analysis. Designed the planned-vs-actual cost model and implemented the supporting finance data pipeline used for reporting and analysis.
Education· 7
Certifications and awards· 20
Hire Andrey Popov today
Post a job and invite Andrey, or contact them directly. No hidden commission.