Senior software engineer · architect · backend / devops
I build systems that keep moving when the load gets real.
Go and Python engineer with 10+ years across fintech, healthcare, city infrastructure and event-driven platforms. I turn ambiguous, high-stakes problems into observable systems teams can operate.
● simulated system traffic
Selected systems
Four contexts where architecture had to survive real-world complexity: city traffic, medical payments, event streams and product analytics.
Led architecture and delivery of a Moscow traffic-control pilot: from team and stack selection to deployment in selected city areas. The three-year system journey combined Go, PostgreSQL, NATS, Kubernetes and ML.
Architected backend and DevOps foundations for US Revenue Cycle Management: payment posting, claim submission, denial management and security-first PHI de-identification for analytics.
Designed a reporting platform for 100M+ daily log records and an identity-provisioning service integrating Box, Salesforce, Slack, Sumo Logic, Splunk and other providers.
Built microservices and reliability practices for product analytics inside a remote bank serving more than 30 million customers — architecture decisions, ADR/RFC, profiling, CI/CD and observability.
Measured results
The common shape across the work: design the system, make it observable, automate delivery, and keep the operational model understandable.
Built PHI de-identification tools for analytics while preserving data utility in US healthcare workflows.
Designed a report-log platform with Sumo Logic, Splunk and other provider integrations.
Developed high-load product analytics services for a remote bank at national scale.
Built Go services for certificate lifecycle automation, WHOIS proxying, monitoring and blacklist checks.
Applied ML in intelligent traffic control research and app classification / clustering pipelines.
Delivered architecture and backend work across regulated, high-load and confidential domains.
Domains & NDA work
Many projects cannot be described in full detail, but the domains matter: regulation, integrations, reliability, legacy constraints and real operational cost.
Consulted and designed systems for Moscow Traffic Center and Moscow Metro contexts: traffic-control architecture, integration boundaries, operational constraints and delivery planning.
Built or explored YouTube-like streaming, crypto/fiat gateways, crypto payments, HFT arbitrage mechanics and an indoor AdTech device concept.
A large part of the work is confidential: backend architecture, integrations, DevOps, observability and reliability work where the domain matters as much as the code.
Experience
A compressed timeline. The common thread is ownership: architecture, delivery, observability and the conversations that keep them aligned.
Anagram · USA · Remote
High-performance medical billing, PHI de-identification, observability and infrastructure automation.
Smartbrain.io · Japan · Remote
Identity provisioning and 100M+ daily report-log architecture with multi-provider integrations.
Tinkoff Bank
High-load product analytics, microservices, profiling, high availability and CI/CD.
Moscow Traffic Center · QuantumSoft
Intelligent traffic-control pilot, green-finance platform and AI-driven talent products for international teams.
PropellerAds · BeGet · RBC · Indusoft
High-load advertising, hosting infrastructure, anti-spam services, ETL/ML and industrial automation.
Open source
Public repositories focused on Go, messaging and application design.
A unified client combining NATS request-reply with streaming delivery guarantees.
02A design-focused test task with a clear application structure and maintainable implementation.
03Concurrent Bitcoin price monitoring with explicit architecture, Docker and PostgreSQL.
Talk
HighLoad++ 2024 talk about building a new intelligent traffic management system for Moscow: Go, PostgreSQL, NATS, Kubernetes, ML and the path from zero to pilot.
PhD results
Candidate thesis work on intelligent control of high-intensity traffic flows: formal models, architecture, synchronization algorithms, RL strategy generation and real pilot validation.
A problem-oriented microservice structure with an RL/ML block, unified controller protocol, shared data bus and scenario knowledge base.
Moscow and New Moscow pilot: 100 controlled traffic-light objects, 24/7 operation across 720 days, zero incidents related to the system.
A decentralized phase-synchronization method for coordinated routes, normal operation and emergency modes with proven convergence assumptions.
Universal driver layer for a heterogeneous road-controller fleet: 8 controller protocols and 16 controller manufacturers in the implementation scope.
A two-loop RL/DQN method with hourly strategy distillation reduced average stops in simulations on real Moscow intersections.
Defense materials list 7 scientific publications, including IEEE and VAK journals, plus 3 state software registrations.
Research & speaking
Traffic-control research, two degrees with honors, ongoing doctoral work and a HighLoad++ talk about taking a complex city system from zero to pilot.
Research paper · SPbPU
Talk · HighLoad++ Moscow
Research paper · SPbPU
IEEE conference paper
MSc, Embedded Intelligent Control Systems · GPA 4.9/5 · BSc, Automation and Management · GPA 5/5
Traffic control lab
A small playable model of the traffic-control problem: demand patterns shift, vehicles queue on red, and every phase decision changes throughput.
Contact
I am interested in architecture, backend and platform challenges where reliability matters and the problem is worth understanding deeply.