System Analyst · AI Integration
Yehor Feshchenko
I turn vague business requests into work a team can build, and I run my own multi-agent AI system every day.
- now
- Senior System Analyst · Lazerhouse
- analyst since
- Sep 2023
- based
- Kyiv, Ukraine · remote
How I work
A made-up request, typical of the real ones. Step through what happens to it.
- heard
- asked
- checked
- written
- ready
“The sales report is slow and sometimes wrong. Can you fix it?”a one-line request from the sales team, Monday morning
1 · Questions I ask
- ❓Which report, and who opens it? The weekly report by region, used by 12 managers.
- ❓How slow? About 40 seconds to open.
- ❓Wrong how? Some returns are missing when you filter by week.
2 · What I check in the system
- 🔎Reproduced it. A return booked on a Sunday lands in the next week.
- 🔎The 40 seconds come from recalculating all history every time the report opens.
3 · The task I write
- Problem
- The weekly report counts Sunday returns in the wrong week, and it takes about 40 seconds to open.
- Reproduce
- Book a return on a Sunday, open the report for that week, and the return is missing.
- Accept when
-
Given a return booked on a Sunday, when I open that week, then it is counted in that week.
Given the report is opened, then it loads in under 5 seconds.
- Priority
- P2: wrong numbers in a report people use for decisions.
✅ Ready for development. No questions left.
Now
- Building acceptance criteria for AI features, and the guardrails that make an AI agent safe to rely on
- Running a personal multi-agent system that routes tasks between AI models and checks their work
Career
Each block is a role, drawn to scale in time. Pick one to read what I did there.
2020
2021
2022
2023
2024
2025
2026
Technical Support Manager
Samsung Electronics Ukraine · Mar 2020 – Jun 2022
- Diagnosed issues from vague user reports; escalated to developers with repro steps and impact
- Wrote troubleshooting guides from recurring cases, moving repeat issues to self-serve resolution
- Technical pre-sales and live-chat support for the online business unit
- Turned recurring issues and chat-quality patterns into feedback for scripts and materials
Technical Account Manager
Lazerhouse · Jul 2022 – Aug 2023
- Support and account management for the systems I later analysed
- Owned work accounts and service access, which later became the access-management process
Middle System Analyst
Lazerhouse · Sep 2023 – Mar 2026
- Turned vague business requests into specified, prioritised development tasks
- Translated requirements between business stakeholders and developers in both directions
- Analysed incidents across CRM, 1C, website and payment integrations and documented each case
Senior System Analyst
Lazerhouse · Apr 2026 – Present
- Own requirements end to end, from elicitation and scoping to dev handoff and UAT
- Designed and own the access-management process for 9 internal systems (request, approval, provisioning, revocation)
- Analyse incidents across CRM, 1C, website and payment integrations and turn recurring ones into process changes
Skills ↔ work
Pick a skill to see where I used it. Pick it again to clear.
Analysis
UML
Data modeling
Scrum
UAT
Tools
REST APIs
Postman
SQL
AI
Retrieval-Augmented Generation (RAG)
Also
- languages
- English · B2+ · Ukrainian · native · Russian · native
- education
- BA, Philology (Slavic Languages and Literatures, incl. Translation), Taras Shevchenko National University of Kyiv, Jun 2021