Pytest Course: Python Test Automation & GitHub Actions CI/CD
Learn Pytest Framework: Python Automation Testing, Unit Testing, API Testing & Test Automation with GitHub Actions CI/CD
Browse all courses
AI & software engineering
Hi, I am
Artem Istranin.
I am an AI engineer, software developer, and technical lead based in Berlin—building products that people want and scale well.
BUILDProducts people want
SCALESystems that last
LEADTeams that deliver
$ current_focus
status: active
Work in the real world
What I build
AI systems, the engineering around them, and teams that can keep improving both.
Over the last several years I have built computer vision models and algorithms for cars and 3D spatial analysis, time-series models for signals that most people never see, and LLM agents for supplier search.
The model is only part of the work. I build the APIs, pipelines, testing, CI/CD, MLOps, LLMOps, and AgentOps around it—the quiet infrastructure that turns an experiment into something people can depend on.
01 / AI
Computer vision, time-series systems, and production LLM agents
02 / SYSTEMS
APIs, data pipelines, cloud infrastructure, and reliable delivery
03 / TEAMS
Technical leadership, mentoring, shared engineering taste, and pragmatic standards
Engineering with people
My Work
Useful systems need more than one clever trick.
I have spent a lot of time leading teams, mentoring engineers, and teaching. That matters because good engineering is also about building shared taste: simple code, useful tests, honest trade-offs, and systems that a tired person can still understand on a Thursday evening.
I co-founded Isometric AI and NestFlow. Those projects pulled me deeper into the messy edge where AI meets real users, uncertain requirements, limited budgets, and production deadlines.
Learn from the work
Read. Practice. Ship.
AI Coding Agents: Production Reliability Matters
AI coding agents make implementation cheap, but reliable systems still require contracts, TDD, layered tests, disciplined reviews, and useful documentation. Here is a production-minded workflow you can apply with any agent.
Browse all field notes