Александр Коленько

I’m a Middle Python Developer with 2+ years of production experience building LLM-powered assistants at T-Bank AI Center, one of the largest fintech AI teams in CIS. I’ve grown from Intern to Middle within 24 months across three shipped products, owning the full feature lifecycle from RFC to production metrics.

I specialize in LLM integration, platform architecture, and async backend systems. I designed and implemented an LLM router to classify requests and route them to specialist assistants, establishing a unified General Assistant entry point for tens of millions of users.

On the General Assistant backend, I deployed from scratch an infrastructure stack with Kafka, Kubernetes, asyncio, and observability, including response streaming and web search integration. My work delivered business impact: +24% user engagement, +1 pp conversion, and −37% incident diagnosis time.

Across my earlier roles, I improved product performance and stability through data-driven iteration and automation—e.g., +1 pp conversion to order for Travel Assistant, and +24.47% sessions per user with reduced incident diagnosis time for the children’s assistant. I also built end-to-end observability (Prometheus, Grafana, Langfuse) and migrated assistants onto an LLM stack (LangChain, LangGraph, Langfuse) to scale orchestration and tracing.