Senior Backend Engineer at T-Bank
Kazakhstan
Kirill Pechurin
Council Member
Kirill Pechurin is a Senior Backend Engineer with over four years of experience building and scaling Python backend systems in high-load production environments. His work spans infrastructure automation, data-intensive analytics platforms, asynchronous processing pipelines, and LLM-powered production solutions, with a consistent focus on clean architecture, measurable performance improvements, and strong testing culture.
Kirill Pechurin currently serves as a Senior Backend Engineer at T-Bank (via its Kazakhstan subsidiary Arctera), a commercial bank with over 10,000 IT staff, where he has worked since July 2024. In this role he designed and implemented a unified IT asset management system from scratch, tracking hardware components across more than 10,000 production servers and warehouse inventory. He led three major version upgrades of an open-source infrastructure management platform serving as the Single Source of Truth for all physical and virtual assets across the organization, a product with over 3,000 internal users spanning 10 repositories, introducing a Continuous Integration mechanism that eliminated reliability uncertainty and raised update confidence to effectively 100%. He expanded enterprise IP address management coverage by 25% by integrating support for five new hardware vendors, improving infrastructure data accuracy across the organization. He also optimized large-scale resource utilization and cost calculations affecting all 10,000 employees by adopting more efficient data structures, reducing processing time by approximately 20% for batch analytics workloads. Additionally, he built and led development of an LLM-powered automation system for procurement and equipment control using the OpenAI and Anthropic APIs, reducing manual operations by 30% across high-frequency decision workflows, and increased test coverage of data collection services from 30% to 80%, directly improving release reliability.
Prior to this, Kirill Pechurin worked as Lead Python Developer at Consul Group, a real estate and construction legal services company, from January 2023 to June 2024. There he developed and integrated three web services into the product ecosystem, including the core user-facing service, extending system capabilities end-to-end. His most significant technical contribution was designing and implementing a partial document update mechanism for Elasticsearch, replacing the standard approach of rewriting entire documents on any change. Working with a corpus of over one million legal and property documents, where each Elasticsearch record combined multiple domain schemas, he restructured storage and update logic at the schema level so that only changed segments were written. This reduced query latency from 5,000ms to 500ms, a tenfold improvement, and the solution sustained subsequent growth in system load without degradation. He also tackled a non-standard integration task, parsing complex XML documents into JSON using XSLT transformations, a tool rarely encountered in conventional Python backend work, which required developing a deep understanding of declarative XSLT logic to build a correct and maintainable mapping. He led a major database schema refactoring in production, preserving existing data while improving schema clarity and future extensibility, and introduced standardized automated testing practices using unittest and pytest across the project, establishing a testing baseline where none had previously existed.
Kirill Pechurin currently serves as a Senior Backend Engineer at T-Bank (via its Kazakhstan subsidiary Arctera), a commercial bank with over 10,000 IT staff, where he has worked since July 2024. In this role he designed and implemented a unified IT asset management system from scratch, tracking hardware components across more than 10,000 production servers and warehouse inventory. He led three major version upgrades of an open-source infrastructure management platform serving as the Single Source of Truth for all physical and virtual assets across the organization, a product with over 3,000 internal users spanning 10 repositories, introducing a Continuous Integration mechanism that eliminated reliability uncertainty and raised update confidence to effectively 100%. He expanded enterprise IP address management coverage by 25% by integrating support for five new hardware vendors, improving infrastructure data accuracy across the organization. He also optimized large-scale resource utilization and cost calculations affecting all 10,000 employees by adopting more efficient data structures, reducing processing time by approximately 20% for batch analytics workloads. Additionally, he built and led development of an LLM-powered automation system for procurement and equipment control using the OpenAI and Anthropic APIs, reducing manual operations by 30% across high-frequency decision workflows, and increased test coverage of data collection services from 30% to 80%, directly improving release reliability.
Prior to this, Kirill Pechurin worked as Lead Python Developer at Consul Group, a real estate and construction legal services company, from January 2023 to June 2024. There he developed and integrated three web services into the product ecosystem, including the core user-facing service, extending system capabilities end-to-end. His most significant technical contribution was designing and implementing a partial document update mechanism for Elasticsearch, replacing the standard approach of rewriting entire documents on any change. Working with a corpus of over one million legal and property documents, where each Elasticsearch record combined multiple domain schemas, he restructured storage and update logic at the schema level so that only changed segments were written. This reduced query latency from 5,000ms to 500ms, a tenfold improvement, and the solution sustained subsequent growth in system load without degradation. He also tackled a non-standard integration task, parsing complex XML documents into JSON using XSLT transformations, a tool rarely encountered in conventional Python backend work, which required developing a deep understanding of declarative XSLT logic to build a correct and maintainable mapping. He led a major database schema refactoring in production, preserving existing data while improving schema clarity and future extensibility, and introduced standardized automated testing practices using unittest and pytest across the project, establishing a testing baseline where none had previously existed.
Earlier in his career, Kirill Pechurin worked as a Python Developer at Digital Spectr, a custom software development firm, from September 2021 to December 2022. Working within a five-person team, he built and containerized two production backend services using Django REST Framework and Docker, establishing CI/CD pipelines from scratch. He implemented full-text search across more than 200,000 objects using Elasticsearch, designing a flexible query-building system that significantly improved search relevance. He also designed core database schemas and introduced unit testing to a previously untested codebase.
Outside his primary roles, Kirill Pechurin has contributed to the open-source community through two public projects. In July 2026 he released structured-eval, a field-level evaluation framework for LLM structured outputs such as JSON and YAML. Built from direct production experience integrating large language models, the framework addresses the absence of tooling for systematically evaluating not just the formal validity of structured responses but also logical field consistency, value accuracy, and hallucination detection. It features a multi-layer scoring system, built-in evaluation mode presets, and integrations with tools including deepeval and LangSmith. He also authored elasticsearch-query-builder, a library designed to simplify the construction of Elasticsearch query filters, developed during his time at Consul Group.
Kirill Pechurin's technical expertise spans Python and SQL as his primary languages, with frameworks including FastAPI, Django, Django REST Framework, and Flask. He works extensively with PostgreSQL, MongoDB, Redis, and Elasticsearch for data storage, and uses Kafka and Celery for messaging and asynchronous processing. His DevOps experience includes Docker, CI/CD pipeline design, and GCP, and he monitors systems using Grafana, Prometheus, Sentry, and the ELK stack. His AI and LLM work encompasses LangChain, the OpenAI and Anthropic APIs, OpenRouter, and structured LLM evaluation methodologies.
Outside his primary roles, Kirill Pechurin has contributed to the open-source community through two public projects. In July 2026 he released structured-eval, a field-level evaluation framework for LLM structured outputs such as JSON and YAML. Built from direct production experience integrating large language models, the framework addresses the absence of tooling for systematically evaluating not just the formal validity of structured responses but also logical field consistency, value accuracy, and hallucination detection. It features a multi-layer scoring system, built-in evaluation mode presets, and integrations with tools including deepeval and LangSmith. He also authored elasticsearch-query-builder, a library designed to simplify the construction of Elasticsearch query filters, developed during his time at Consul Group.
Kirill Pechurin's technical expertise spans Python and SQL as his primary languages, with frameworks including FastAPI, Django, Django REST Framework, and Flask. He works extensively with PostgreSQL, MongoDB, Redis, and Elasticsearch for data storage, and uses Kafka and Celery for messaging and asynchronous processing. His DevOps experience includes Docker, CI/CD pipeline design, and GCP, and he monitors systems using Grafana, Prometheus, Sentry, and the ELK stack. His AI and LLM work encompasses LangChain, the OpenAI and Anthropic APIs, OpenRouter, and structured LLM evaluation methodologies.