Ilya Strelov.
Senior Software Engineer in TypeScript/Node.js, Go, and Python (Rust in progress), building distributed systems at scale — microservices on Kubernetes, event-driven pipelines on Kafka, latency-sensitive APIs, infrastructure-as-code by default. Also building production LLM systems: multi-agent code review across dozens of repositories and agentic search over large datasets. I own each feature from architecture through deployment and observability.
- Owned 21 microservices at ~4K RPS each (84K RPS aggregate) inside a 150+ service Kubernetes cluster, holding a 99.999% availability SLA for a 10M+ user enterprise messaging platform.
- Led the Threads feature end-to-end — architect-validated design docs, modifying 5 microservices while preserving backward compatibility under full production load.
- Built an LLM multi-agent code-review pipeline adopted voluntarily by 150+ engineers across 60 repositories, posting inline comments on security, performance, and architecture against internal coding guidelines.
- Built an automated vulnerability-remediation pipeline across 21 repositories: a scanner files context-rich Jira tickets and an LLM agent applies fixes and opens PRs for human review.
- Co-built a custom request throttler in Go and Redis between auth and downstream services, smoothing peak spikes with graceful degradation instead of hard rejects.
- Ran on-call with SRE-style ownership: SLOs and error budgets, alerting and runbooks, blameless postmortems, and Prometheus/Grafana/tracing dashboards.
- Led the migration to a serverless architecture (NestJS on AWS Lambda), cutting maintenance and running costs by 40% while improving availability.
- Owned architecture decisions as Tech Lead, setting service boundaries and technical direction across the team.
- Unblocked new-customer onboarding by building an in-house emulation of a dependent team's integration, letting us onboard and gather product feedback ahead of the upstream work.
- Built full-stack features end to end (React, Next.js, DynamoDB, RDS PostgreSQL); managed IaC with Terraform and AWS CDK.
- Ran technical interviews and hiring, and mentored peers through code review and quality initiatives.
- Bootstrapped a new product line from zero inside the bank — architecture, infrastructure (Docker, OpenShift), CI/CD (Jenkins), and observability within a large corporate stack.
- Cut report loading time by 95% by profiling the reporting backend (Node.js, NestJS) and reworking its slowest data and query paths.
- Built the fintech dashboard in Next.js and React, owning component architecture and the rendering work that sped up massive reports by 97%.
- Hired and onboarded the engineering team for the new product line and ran technical interviews throughout.
- Led the migration of a monolith to a Kubernetes microservice architecture, designing service boundaries and orchestrating the phased rollout.
- Built and scaled backend services across PHP, Go, Python, and Node.js, picking the language per service to match its workload.
- Established the company's first observability and on-call practice — Prometheus, Grafana, and alerting from scratch.
- Grew from Software Developer to Principal, owning architectural decisions, code-review standards, hiring, and mentoring.
Built from scratch — infrastructure, CI/CD, observability, product. A pipeline of 30+ specialized AI agents analyzing GitHub PRs for security, performance, and architecture, with false positives reduced by scoping each agent to a narrow responsibility.
Query-centric job-search engine with a CLI: aggregates postings from company portals (Greenhouse, Ashby, Lever) and exposes them through a fast search and tracker-sync workflow.
Bootstrapped from zero. Analytics platform indexing 1.4M+ Telegram channels and 613M+ messages. Agentic search (LLM with 8 tool calls), hybrid search (Meilisearch + pgvector), ClickHouse OLAP.
Bootstrapped end-to-end: architecture, infrastructure, billing, provisioning. Self-service platform that lets non-technical users deploy their own AI Telegram bots in under 3 seconds, handling the full lifecycle from provisioning to billing.