Nikita Kurinnyi
Lead Backend Engineer (Python, AI) · High-Load Systems · Microservices
Summary
- 12+ years as a Python backend engineer focused on high-load, distributed systems (FinTech, GameDev, MarTech, AI).
- Led and scaled backend teams (up to 5 engineers) and delivered microservice architectures end-to-end.
- Designed and optimized real-time data pipelines (Kafka, PySpark, BigQuery) processing 250M+ events/day.
- Build and leverage AI-native workflows and agentic tooling (LangGraph, AutoGen, LlamaIndex, RAG) to speed up delivery and automate engineering work.
Skills
- Languages
- Python, Go, SQL, Bash
- Frameworks
- FastAPI, Django, asyncio, Celery, SQLAlchemy, Alembic, pytest
- Databases & Caching
- PostgreSQL, MySQL, Redis, ClickHouse, BigQuery, DynamoDB, Elasticsearch, MongoDB
- Messaging & Streaming
- Kafka, RabbitMQ, Redis Pub/Sub
- Infrastructure & DevOps
- AWS (EC2, S3, Lambda, SQS), Docker, Kubernetes, Terraform, Linux, Nginx, CI/CD
- Observability
- OpenTelemetry, Jaeger, Grafana, Prometheus, Sentry, ELK
- AI/LLM
- LangGraph, RAG, AutoGen, LlamaIndex, OpenAI API, embeddings
- Other
- gRPC, Protobuf, GraphQL, PySpark, ETL, Git
Languages
- Russian - C2 (Native)
- English - B1
Experience
- Developed an internal operations dashboard that automated investigations of missing payments, reducing manual processing time per case by 90%+ (from ~6 minutes to ~30–40 seconds).
- Architected and built centralized credential management for RPA bots and polling services, improving reliability and adding clear monitoring for failures.
- Designed and implemented a multi-layer data model for analytics (raw → stg → dds → emart), enabling consistent BI dashboards and better data quality.
- Integrated OpenTelemetry distributed tracing to speed up incident response and identify performance bottlenecks (e.g., GraphQL N+1 patterns).
- Championed a migration initiative from a legacy GraphQL stack to a REST-based architecture with FastAPI to improve maintainability and developer velocity.
- Co-designed and led the development of the core backend architecture: 4 foundational microservices (Auth, User, Wallet, Transactions) using Python and gRPC/Protobuf.
- Engineered fault-tolerant payment integrations with Stripe and the National Bank of Kuwait, implementing robust idempotency to guarantee transaction integrity.
- Established engineering foundations (CI/CD, code review standards, stack/tooling choices) and participated in backend hiring.
- Conducted load testing with Yandex.Tank and tuned services for stable performance at 100 RPS baseline.
- Implemented end-to-end observability (OpenTelemetry, Jaeger, Grafana, Sentry) to improve cross-service visibility and debugging speed.
- Hired as the first backend engineer to build the cloud infrastructure from scratch on AWS for an AI-powered chat companion application.
- Collaborated with the ML team to optimize the AI response pipeline (~10 model calls), reducing p95 latency by 50%+ (from ~5s to ~2.2s).
- Architected the data layer using DynamoDB for low-latency chat history and embedding storage, and BigQuery for scalable analytics.
- Designed and implemented a high-throughput, real-time analytics pipeline (FastAPI, Kafka, PySpark, BigQuery) processing 250M+ events/day.
- Delivered and maintained critical hourly data aggregates for reporting and product analytics.
- Led a backend team: sprint planning, technical roadmap discussions, and code reviews to ensure predictable delivery.
- Led backend development for Tochka XS, a highly automated SMB banking product designed for near zero-touch support.
- Improved reliability and observability by introducing test coverage and production monitoring (Prometheus) plus error tracking (Sentry) in a legacy system.
- Developed and maintained core services using Python/asyncio and PostgreSQL, handling complex integrations with internal banking systems.
- Developed a high-performance CS:GO replay parser to feed data into ML training and analytics pipelines.
- Maintained reliable real-time Dota 2 data feeds used for live tournament broadcasts.
- Built third-party API integrations and a notification system for end users.
- Contributed to gradually decomposing a monolithic Django app into services using Celery and Redis.
- Built and shipped backend services for startups and freelance clients (Django, databases, caching, message queues).