Full-stack engineer · Moscow · Since 2022

Software that doesn’t break. Oleg Bozhenov

Production backends for fintech, LLM platforms and consumer apps — built in Python on FastAPI, deployed across Docker, Postgres, Kafka and Redis.

Start a project→ Selected work↓
BasedMoscow, Russia
Core stackPython · FastAPI · Postgres · Kafka
ShippedETL · LLM · Fintech · VPN · Voice AI
§ 01 — SERVICES

What I build for clients

01 · Backend systems

APIs, microservices and event-driven platforms

Greenfield and brownfield backends on Python/FastAPI. Clean service boundaries, async workers, production-grade auth and observability from day one.

FastAPIPostgreSQLRedisKeycloak
02 · Data pipelines

ETL, DWH and streaming at scale

Airflow orchestration, Kafka / RabbitMQ ingestion, ClickHouse analytics. Credit scoring, change-data-capture, and legacy migrations that don't lose rows.

AirflowKafkaClickHouseCDC
03 · LLM & AI plumbing

Production integrations, not demos

OpenAI, GigaChat and retrieval pipelines wired into real products — chat, search, ranking, voice. BM25 + embeddings + LambdaMART when the answer has to be good.

OpenAIGigaChatLightGBMRetrieval
§ 02 — PROFILE

A short transmission

Backend engineer with three years of production experience shipping API-driven systems, microservices and infrastructure-heavy platforms — credit scoring, LLM products, VPN, ETL at scale.

I care about the parts of backend work that rarely make it into slide decks: clear contracts between services, failure modes that are obvious before they happen, and observability that turns incidents into short stories instead of mysteries.

FocusBackend · Distributed systems
StackPython · FastAPI · Postgres
InfraDocker · Kafka · Redis · Nginx
CloudAWS · Yandex Cloud
BasedMoscow, Russia · UTC+3
LanguagesEnglish · Russian
§ 03 — SELECTED WORK

Things I've shipped

P · 012025 → present · NDA

Document Intelligence Platform

Enterprise · closed contour · on-prem LLM

Backend for an enterprise document workbench deployed in a closed environment — ChatRAG over internal corpora, an LLM-driven Checker for style, formatting and standards compliance, and a document editor with an agentic AI assistant.

  • ChatRAG retrieval over document corpora with on-prem inference
  • LLM Checker — style, formatting and standards compliance
  • Editor with an AI agent for structured document editing
  • Runs fully inside a closed contour — no external network
PythonFastAPIPostgreSQLLLMRAGAgentsOn-prem
P · 022026 → present

LegalTech Defense Brief Generator

Turnkey full-stack build · OCR → LLM pipeline · penalty reduction

End-to-end product built solo — React frontend, Python backend, and the full OCR → LLM pipeline. Users upload a lawsuit against a debtor along with sales contracts and supporting documents; Yandex AI OCR extracts text; OpenAI assembles a legally-reasoned counter-response aimed at reducing penalties and fines.

  • Shipped solo, turnkey — React frontend, FastAPI backend, infra
  • Document ingest — scanned lawsuits, sales contracts, supporting evidence
  • Yandex AI OCR for high-fidelity text extraction
  • OpenAI prompt-chain composes a structured, legally-grounded response
  • Optimized to reduce penalties and fines in the debtor's reply
ReactTypeScriptPythonFastAPIPostgreSQLYandex AI OCROpenAIDocker
P · 032024 — 2025

Educational LLM Platform

Large-scale learning product · internal Big-Tech client

Backend microservices and REST APIs for an LLM-powered educational platform with Telegram integrations, Keycloak-based auth and multi-provider LLM routing.

  • Authentication and authorization via Keycloak
  • Service ↔ Telegram bot APIs
  • OpenAI + GigaChat routing for chat, search, summarization
PythonFastAPIPostgreSQLTortoise ORMRedisDockerKeycloak
P · 042025

LegalTech Retrieval Engine

Bachelor thesis · patent search & ranking

Backend combining classical retrieval with learning-to-rank for patent search and natural-language query response.

  • BM25 + embedding retrieval
  • LambdaMART ranking
  • Query-processing and response API endpoints
PythonFastAPIPostgreSQLOpenAILightGBM
P · 052025

AI Voice Platform

Telephony integration · scoring pipelines

Backend for an AI-based voice communication platform integrating third-party telephony APIs, onboarding flows and scoring pipelines.

  • Third-party telephony API integration
  • User onboarding + data persistence
  • Scoring pipelines
PythonFastAPIPostgreSQLDockerRetell APIOpenAI
P · 062025 → present

Production VPN Service

Independent product · Xray core · native macOS & iOS apps

A self-operated production VPN — Xray-based backend with subscription lifecycle, concurrent-session abuse protection, containerized deploy, and native client apps for macOS and iOS.

  • Xray core with multi-protocol inbound/outbound configuration
  • Concurrent-session abuse protection
  • Subscription + link-generation APIs
  • Telegram bot for payments and lifecycle
  • Native macOS & iOS apps — Swift, NetworkExtension, seamless config import
  • Nginx + SSL automation + container deploy
PythonFastAPIXrayDockerNginxSwiftmacOSiOS
P · 072024

AI Personal-Growth Coach

Conversational LLM product · branching methodologies

An AI coach backend for personal growth — delivers structured coaching methodologies while branching the conversation in real time based on the user's intent and responses.

  • Intent classification on every user turn
  • Branching dialogue graph over a library of coaching methodologies
  • State persistence across long-running coaching sessions
  • LLM-guided session summaries and next-step recommendations
PythonFastAPIPostgreSQLLLMIntentDialog graph
§ 04 — EXPERIENCE

Operational log

2022 — 2024
2024 — Present Python Backend Developer

WB Digital

Moscow, Russia

Data-heavy backend services and ETL orchestration for a fintech-adjacent platform. Ownership from pipeline to production monitoring.

  • Designed and maintained backend services and ETL pipelines in Python + Apache Airflow
  • Built data-intensive services deployed via Docker
  • Implemented credit-scoring backend; migrated legacy logic from KNIME to Python (PyPMML)
  • Integrated monitoring and logging with Prometheus and Graylog
  • RabbitMQ + Kafka for event-driven ingestion and DWH aggregation
Mar — Jun 2024 Business Analyst · backend-oriented

Tinkoff Bank

Moscow, Russia

Bridged engineering and product on internal backend systems — specifications, bottleneck analysis, delivery.

  • Documented backend and business processes for internal systems
  • Collaborated with engineering on task decomposition and delivery
  • Identified system bottlenecks, vulnerabilities and logic errors
  • Analyzed user-interaction data to improve backend workflows
2022 — 2023 Teaching Assistant · CS fundamentals

HSE University

Faculty of Computer Science

Supporting first-year mathematical foundations coursework in one of Russia's strongest CS faculties.

  • Assisted in teaching linear algebra and discrete mathematics
  • Evaluated assignments and examinations
  • Conducted technical consultations for students
§ 05 — EDUCATION

Formal training

Sep 2025 — Jul 2027

MSc — Financial Technologies & Data Analysis

HSE University · Faculty of Computer Science
Moscow, RU
Sep 2021 — Jul 2025

BSc — Applied Data Analysis

HSE University · Faculty of Computer Science
Moscow, RU
Oct 2022 — May 2025

BSc — Data Science & Business Analytics

London School of Economics
London, UK
§ 06 — STACK

Shipped in production

Languages 04
  • Python primary
  • C++ systems
  • Bash ops
  • Swift side
Backend 05
  • FastAPI core
  • Django
  • Flask
  • REST APIs
  • Microservices
Data 06
  • PostgreSQL primary
  • ClickHouse olap
  • Redis cache
  • ETL / ELT
  • DWH
  • CDC kafka connect
Platform 09
  • Docker
  • Kafka
  • RabbitMQ
  • Nginx
  • Prometheus
  • Grafana
  • Graylog
  • Keycloak
  • AWS · YC cloud
— Let's work together

Let's ship something that stays up.

If you need a backend that can carry real traffic, a data pipeline that doesn't lose rows, or an LLM integration that isn't a demo — I'm open to senior contract and full-time engagements.

Start a conversation→
§ 07 — CONTACT

Open a channel

Say hello —
[email protected]

The fastest way in. I read everything and reply within a day.

Email[email protected]↗ Telegram@obozhenov↗ GitHub@ducklingsam↗ LinkedIn/in/odbozhenov↗

TWEAKS //cfg