Frontend Engineer, Full Stack Engineer
About the specialist
Full-stack engineer working mainly in TypeScript and React, with Python and gRPC in production. Sole engineer on a distributed cache for ML training data - I built the React dashboard, the Python SDK on PyPI and the gRPC backend while shipping a Next.js 16 storefront rewrite in parallel. Practical depth in ordering, consistency and partial failure: the things the front end usually hides.
Skills
Work experience
[скрыто] Founding Full Stack Engineer at Hord AI
React, TypeScript, Zustand, Tailwind CSS, Python, gRPC, PostgreSQL, AWS, Docker, Grafana
Sole engineer on a distributed caching platform for ML training data - owned the React dashboard,
Python SDK, gRPC backend and production AWS infrastructure end to end, from architecture through
deploys and on-call
Shipped a React/TypeScript dashboard (7 pages, 34 reusable components, Zustand, gRPC-Web via Envoy) with live cluster telemetry, embedded Grafana panels, and cache-hit / GPU-starvation / p50–p99 latency visualisations
Authored and released the client SDK to PyPI (Python 3.8–3.11, actively maintained): a peer-to-peer disk cache between PyTorch/TF/JAX jobs and S3-compatible storage - consistent-hash placement, LMDB
store with LRU eviction and pinning, HTTP range reads
Cut warm-cache training epochs 6.5× and eliminated all S3 requests after the first pull, benchmarked on a 2× A5000 GPU cluster
Designed a gRPC control plane (5 services, ~30 RPCs) in async Python: JWT auth, cluster and API-keymanagement, streaming metrics ingest, server-streamed live updates, and Stripe billing
Ran production infrastructure: 2× EC2 behind an ALB, Multi-AZ RDS PostgreSQL 17, deploys,
and GitHub Actions CI across frontend, backend, SDK and Alembic migration round-trips, with automated publishing to PyPI
Wrote ~340 tests across 33 files, including Hypothesis fuzz tests on the wire decoder
[скрыто] Frontend Developer at Hepteract Software Solutions and APIQ Media Technologies
Vue.js, Nuxt.js 2/3, TypeScript, Tailwind CSS
Developed a multi-role SaaS platform powering 100K+ streamers, 10K+ Twitch creators and 80+ brands across 400+ ad campaigns, with dedicated dashboards for streamers, advertisers, agencies and administrators
Led API layer refactoring spanning ~140 endpoints across 6 role modules: unified integration patterns into a single declarative config (path/method/data tuples), eliminated duplicated request logic, and accelerated onboarding for new developers
Implemented financial analytics, profile management, and campaign management interfaces, reducing
reliance on third-party analytics tools
Developed Twitch extensions and rebuilt the company marketing website (apiq.live) with custom
animations and responsive layouts
Education
Bachelor’s degree at Saint Petersburg State University Environmental Engineering, [скрыто] Professional retraining at Yandex School of Data Analysis Web developer, 2022-2023
Preferred geography
Serbia, ЕС
Languages
English B2, Russian native