For a tech lead
Stack, architecture, code.
Here's how I design and what I write by hand: from pre-sale to production — REST, event pipelines, Kubernetes.
- My core is Java/Kotlin + Spring + Kafka; I close full-stack tasks with React.
- Behind me: Sber's high-load, 20+ microservices at TextBack, and four years at the Reckue studio.
- I use AI agents seriously, but I design the architecture myself and keep the code under control.
Core
- Java 8–21, Kotlin, Spring Boot and Spring Security, Hibernate, jOOQ.
- I design REST and WebSocket APIs; at Sber I built gRPC integrations.
High-load at Sber
- Three years on SberSpasibo: the platform core, the mobile app and the in-bank section — millions of users.
- I carried out a platform migration with zero downtime in production.
- I built the error-handling panel for the Aeroflot bonus program — full stack, faster incident response.
Data and events
- PostgreSQL and Apache Kafka at the core; MongoDB, Redis and WebSocket alongside.
- I debugged asynchronous services on ActiveMQ in the TextBack messaging platform.
Microservices and legacy
- TextBack: 20+ microservices — chats, media files, messenger integrations.
- I fixed critical bugs in media processing and database access, and refactored large legacy classes.
Delivery
- Docker, Kubernetes, Nginx, GitLab CI/CD, Jenkins, Keycloak.
- At Sber I built CI/CD on Jenkins + Docker + Kubernetes.
Front end, when needed
- React, Redux, TypeScript, NestJS — I wrote Sber's admin panels and my studio projects' front ends myself.
AI at work
- I do agentic development in Claude Code: my own MCP servers, integrations with Jira, Confluence and Telegram.
- I train MLP/NLP classifiers for applied tasks and use prompt engineering as a working tool.
- I control the AI's output and design the architecture myself — I build large projects fast without losing quality.
- An honest boundary: I don't claim expertise in training large models, ML infrastructure, or deep neural network architecture.
Process and team
- Scrum, retros, code review, one-on-ones, mentoring; teams of 5–20 people.
- At NLab I introduced sprints on large projects — fewer hotfixes, key features shipped faster.
- At pre-sales I defended technical estimates worth over 5M ₽.
A match?
Stack and project details are in the CV. For code questions, message me in Telegram — I reply quickly.