AI Agent Infrastructure Engineer — observability, remote control, and configuration systems for AI coding agents
AI coding agents are a new class of developer, and they require their own ops stack. I build the infrastructure that makes autonomous coding reliable, observable, and controllable. My work lives at the intersection of developer tooling and agentic AI — focusing on the operational layer around frameworks like OpenCode and Oh-My-OpenCode. I build the telemetry, remote control systems, and configuration architectures that turn experimental AI workflows into production-grade pipelines.
Projects
Zero-config local dashboard for OpenCode. Monitors AI coding sessions, token burn, and agent status in real-time by reading directly from the native SQLite database — no instrumentation or API keys required.
Active
Telegram Mini App to monitor and control OpenCode sessions from your phone. Check live status, send prompts, abort stuck tasks, and track token usage without opening your IDE. Self-hosted, privacy-first, built with Hono, React, and TypeScript.
Active
Production-ready OpenCode + Oh-My-OpenCode configuration pack. 7 AI providers, 12 specialized agents, git safety and worktree plugins, plus a wisdom management system. Reusable presets, skills, and setup examples for AI coding workflows. One-command install with automatic backups.
https://github.com/EZotoff/ez-omo-config/stargazers
Active
Web-based patient decision aid that calculates personalized risk/benefit profiles for hip and knee replacement surgeries, powered by National Joint Registry data. Built at the University of Sheffield (ScHARR).
Completed
A generative presentation toolkit that transforms text prompts into complete, visually structured slide decks, handling everything from narrative flow to layout design.
Active
An AI agent simulation platform where agents wake up in a Docker sandbox with 1,000 credits. To survive, they must process files and navigate a credit economy using 8 built-in tools, while hidden validators test their true capabilities.
Active
Automated intelligence pipeline for niche markets. Monitors industry sources, analyzes trends, and synthesizes findings into targeted newsletters and multi-channel alerts.
Active
Skills
Work with Evgeny
Advisory
I help teams get AI coding agents into production — choosing the right provider stack, building reliable workflows, adding observability, and creating the custom tooling that makes agents dependable at scale. I also audit and fix existing setups for teams already using agents but hitting reliability or debugging walls. Typical engagements: production coding agent rollout, custom agent tooling, or infrastructure audit and remediation.
Custom
Contract
I design and build non-coding AI agent systems for teams that need AI beyond the IDE — research automation, content synthesis, decision-support pipelines, and multi-agent architectures for specific business processes. If you have a repetitive knowledge-work bottleneck that LLMs could handle but nobody has built the pipeline yet, that's my lane.
Custom
Press
Frontiers in Artificial Intelligence · Nov 2021
CycleStyleGAN-Based Knowledge Transfer for a Machining Digital Twin“Digital twin simulation tool that captures the dynamics of a machining vibration signal from a source model and adapts them to a target environment using a novel CycleStyleGAN domain adaptation algorithm based on generative adversarial networks.”
EANN 2020 Conference · Jan 2020
Towards a Digital Twin with Generative Adversarial Network Modelling of Machining VibrationInternational Journal of Medical Informatics · Oct 2020
JointCalc: A web-based personalised patient decision support tool for joint replacement“First comprehensive patient-facing web-based decision support tool for joint replacement. Published in Int'l Journal of Medical Informatics with 16 citations.”
Integrated Computer-Aided Engineering · Aug 2021
Conditional StyleGAN Modelling and Analysis for a Machining Digital TwinExperience
Jan 2023 – Present
Designed and implemented LLM applications and AI systems for government and industrial partners across the finance, manufacturing, and healthcare sectors, with a specific focus on deploying AI in regulated environments.
Jan 2018 – Dec 2020
Engineered JointCalc, a patient decision support tool driven by survival analysis models. This work was subsequently published in the International Journal of Medical Informatics.
Jan 2008 – Dec 2017
Architected data warehouses, European mobile banking systems, and automated testing frameworks. Progressed from developer to Consultant through four promotions over nine years.
Jan 2021 – Dec 2023
Developed manufacturing digital twins and aerospace-grade prediction systems using Generative Adversarial Networks and statistical modeling to simulate machining processes.
Education
PhD Research
PhD research on applied Generative AI for manufacturing. Developed a domain adaptation model that lets newly installed machining tools use existing neural networks immediately, without retraining, while preserving prediction accuracy.