Engineer · Kyiv · systems that are measured, not felt

Bohdan Fesenko

I build the seam between AI agents and people, and then I make it reviewable.

Engineering lead of a five-person team that went from no process to a shipped product, largest single contributor to the codebase while leading it, and the person who made agent-assisted development a reviewable practice there. On my own time: measured detection-and-tracking pipelines and a shared memory for a fleet of agents.

tech leadagent systemsMCPTypeScriptPythonapplied MLtracking and estimation

A moving target is in every one of these frames.Its signal is 3 dB above the noise: a spot slightly, unreliably brighter than its neighbours. You cannot point at it, and neither can a threshold.

Live · runs in your browser

Measured, not staged

Threshold on each frame2 of 100
Track-before-detect, 16 frames85 of 100

Experiment E3 in track-lab: the Python reference, 100 noise draws at 3 dB, one false alarm per frame allowed to both. How it is measured

Featured projectDetection and tracking2026

track-lab: from noisy detections to confirmed tracks, measured

Targets too faint for any single frame, noise that looks like targets, labels that arrive late. A reference tracker and a simulator that knows the truth, with every decision measured and runnable in the page.

  • Simulator with exact truth; Kalman and IMM; gating; GNN and JPDA; M-of-N and SPRT confirmation; track-before-detect; bearing-only fusion
  • Browser simulator checked against the Python reference by a test: pd and pfa to 1e-6, assignment cost equal to scipy's
  • Evaluation module written before the tracker, including MOTA, IDF1 and HOTA and their rank stability under corrupted labels
  • Confirmation frontier measured on 1.3 hours of data per cell: 24 false tracks per hour at 2.2 s lead time for the strictest policy; the amplitude-aided sequential test sits off the frontier, and the page says why
PythonnumpyscipyJavaScriptCanvas

Selected work

All work ->
Agent engineeringDec 2025 – Sep 2026

Agent-assisted development as a reviewable practice

I made AI sessions part of code review, with a rubric, named red flags and an onboarding gate. Velocity without reviewability is debt.

Claude CodeCodexBMADGitHub ActionsNotion APITypeScript
  • Agentic workflow introduced to the company on day one; repo-resident agent commands and a story-sync workflow that made agent-authored story files in git the tracker of record
  • An AI-coding course as onboarding gate: new developers complete it before starting work
  • Review of AI sessions as part of code review, scored on prompting quality, review discipline and workflow efficiency, with named red flags
Platform2026

An MCP surface that refuses to boot if a tool lies about itself

83 tools, five live resources over server-sent events, OAuth with dynamic client registration for third-party clients, and a registry validated at start-up.

83-tool catalog with per-tool annotations and boot-time registry validation

TypeScriptFastifyModel Context ProtocolSSE
Agent engineering2026

An assistant that operates this site in front of you

Ask it something and it flies to the page, points at the section, runs the demo and turns its knobs, then answers with cards built from the site's own content, so it cannot invent a number.

One typed action registry drives four directors: the live model, recorded tours, voice, and any browser agent through WebMCP

TypeScriptClaude APIOpenAI RealtimeWebRTC
Real-time C++2026

Few-pixel targets from a moving camera, in real time on the edge

A target three pixels wide carries almost no information in space. All of it is in time, so the pipeline integrates before it decides.

Registration residual on real 4K moving-camera footage: held-out median 0.080 px over 706 pairs, zero failures, at the instrument floor

C++17pybind11numpyaarch64
Infrastructure2026

One memory for a fleet of coding agents

Several agents, several machines, one shared data layer with mounts, signal streams, addressed whispers and a secrets vault, and a rule about who may write what.

First generation: a single SQLite file exposed as an MCP server with 12 tools; documents with frontmatter, direct and broadcast signals, presence, full-text search; no external services

RustTypeScriptSQLite FTS5PostgreSQL
Product2026

A mobile product shipped to TestFlight in six weeks, by one engineer with agents

A social app for a city's motorcycle community, with a Telegram bot for the chats riders already use: iOS and Android from one codebase, one verification gate shared by people and CI.

From the first commit on 2026-08-10 to a production build on TestFlight on 2026-09-17; 1,235 commits by 2026-09-23

React NativeExpoTypeScriptSupabase

Contact

The most useful first message is not a role title. Tell me where a person like this is needed and what your hardest problem is. I answer within a day.