Сервіс пошуку роботи №1 в Україні
Виталий
Fullstack-програміст
- Розглядає посади:
- Fullstack-програміст, Frontend-програміст, Backend-програміст, AI engineer, Automation engineer
- Вік:
- 39 років
- Місто проживання:
- Житомир
- Готовий працювати:
- Дистанційно
Контактна інформація
Шукач вказав: ТелефонЕл. поштуМесенджер
Прізвище, контакти та світлина доступні тільки для зареєстрованих роботодавців. Щоб отримати доступ до особистих даних кандидатів, увійдіть як роботодавець або зареєструйтеся.
Отримати контакти цього кандидата можна на сторінці https://www.work.ua/resumes/20543923/
Завантажений файл
Версія для швидкого
перегляду
Це резюме розміщено у вигляді файлу. Ця версія для швидкого перегляду може бути гіршою за оригінал резюме.
VITALIY HRYBINYK
Senior Fullstack Engineer | .NET + React | Multi-agent LLM systems
Ukraine (Remote) · [відкрити контакти ](див. вище в блоці «контактна інформація») · [відкрити контакти ](див. вище в блоці «контактна інформація») · [відкрити контакти ](див. вище в блоці «контактна інформація»)
SUMMARY
Fullstack engineer with 7+ years in .NET and React (12 years in software engineering overall), across IoT, eCommerce,
agriculture, Sitecore and security domains. Most recently designed and shipped to production a memory-backed multi-agent
LLM platform: microservices, gRPC agent orchestration, hybrid RAG and its own CI/CD. Strong in distributed backend
architecture, Azure and modern frontend.
CORE SKILLS
Backend: C#, .NET Core / .NET 10, ASP.NET Core, WebAPI, REST API, EF Core, gRPC, SignalR, Node.js, Express.js,
RabbitMQ, Python (working knowledge)
Frontend: React, React Native, JavaScript, TypeScript, Redux, MobX, Electron, Next.js, SCSS/SASS/LESS, Material-UI
AI / LLM: AI agents and multi-agent orchestration, multi-provider LLM runtime (OpenAI, Anthropic, DeepSeek), RAG
(pgvector, Neo4j, cross-encoder reranking), vector embeddings, prompt versioning, Brier calibration, local model serving, MCP
(Model Context Protocol, working knowledge), AI-assisted development workflows
Architecture: microservices, distributed systems, explicit execution DAGs, SOLID, scalability, code review, mentoring,
Agile/Scrum
Data: PostgreSQL, MS SQL, MySQL, SQLite, Redis, Neo4j
Cloud & DevOps: Azure (primary), AWS (working knowledge), Azure Functions, Azure IoT Hub, Docker, CI/CD (GitHub
Actions, Azure DevOps), YARP, health checks and observability
Testing: xUnit, MS Fakes, Vitest, Playwright, Cypress, unit and integration testing, performance profiling
PROFESSIONAL EXPERIENCE
Independent Fullstack Engineer / Contractor — Self-employed
trading_bot — market research system (own product)
09/2026 – Present · TypeScript, Node.js, Vitest
• Built a research system measuring the link between news and price movement: 94 modules, ~26,900 lines, 1,685
automated tests across 77 files (all green).
• Implemented interval arithmetic (an empty intersection triggers refusal) and Dempster-Shafer evidence theory to separate
conflict from ignorance; refusal is a first-class outcome, not a failure.
• Built a blind-testing harness over historical data: 181 decision steps with the full production logic and no look-ahead — the
system correctly refused all 181 times.
• Enforced a mandatory asOf time parameter across all time-dependent calculations to eliminate look-ahead bias, plus
train/holdout separation and clustered confidence intervals.
• Fixed a 502x performance defect (26,125 ms to 52 ms per call) and rejected a 730x speed-up because a parity lock proved
it leaked future data.
• Served an LLM locally (LM Studio) and calibrated outputs by Brier score; measurement showed no directional skill, so the
model was repurposed as a significance measurer.
veriditarot — multi-agent LLM platform (own product, live in production)
02/2026 – 08/2026 · .NET 10, gRPC, React/Next.js, PostgreSQL, Neo4j, Docker
• Designed and shipped a memory-backed multi-agent platform: 19 .NET microservices, 7 gRPC reasoning agents, a
32-container Docker stack and 4 clients (Next.js web + admin, Expo mobile, Electron desktop).
• Built an explicit artefact-DAG execution engine with bounded concurrency (max 3 concurrent LLM calls per session),
per-agent gates, retries and a circuit breaker keyed by service + method + promptKey; a full Celtic Cross reading fans out to
37 LLM calls in total. Live at veriditarot.com.
• Implemented hybrid RAG: 768-dim pgvector embeddings + Neo4j knowledge graph + LLM-based chunk tagging +
cross-encoder reranking, with a reading-memory feedback loop (durable outbox with reconciliation) that personalises later
sessions.
• Added habit/novelty/decay weighting for cross-session memory ranking, replacing static score constants.
• Shipped CI/CD: a GitHub Actions self-hosted runner builds images to GHCR; the VPS pulls and restarts via docker
compose over a pull-only SSH key. Validated with a 25-concurrent-reading load test: 0 errors, 32% peak CPU on 4 vCPU.
Client contracts — freelance engagements
07/2023 – 12/2025 · .NET Core, React, TypeScript, PostgreSQL
• Multiple short-term fullstack engagements (3-5 months each) for various clients.
• Delivered .NET Core backends and React frontends end to end.
Techery — Fullstack Engineer | 01/2023 – 06/2023
• Fullstack development on client web products: .NET backend services and React frontend.
Tech: .NET Core, React, TypeScript
Mapleton Hill (via EXDST) — Fullstack Engineer | 10/2021 – 12/2022
• Sitecore CMS and public website for Mammoth Mountain Resort (mammothmountain.com).
• Agricultural platform (farm management plus satellite navigation for farming machinery): joined as a fullstack engineer to
bridge the frontend team and the .NET backend.
• Rescued platform performance: reproduced the slowdown by throttling the network, then traced it to 30-40 requests per click
— each of four panels plus the map re-fetched the same deeply nested JSON instead of parsing it once, cascading
re-renders across every panel.
• Fixed the backend half: steady ~30 req/s was served through Azure Functions, which are built for event-driven,
scale-to-zero workloads rather than constant low-latency traffic, and for data that in some cases changes only monthly.
Partially moved off them; roughly halved the request volume.
• Designed client-server architecture for IoT device communication.
• Set up and maintained CI/CD pipelines on Azure.
Tech: .NET Core, Sitecore, React, Redux, Azure Functions, Azure Cloud, CI/CD
Varyence — Fullstack Engineer / Tech Lead | 04/2021 – 09/2021
• Led a smart sports equipment platform end to end: Azure IoT Hub ingests telemetry from the training device, the backend
collects and stores it, and the mobile app shows training data.
• Implemented QR-code pairing between the trainer and the user's phone.
• Found that the vendor-supplied architecture used an Azure IoT tier unfit for production multi-device scale; redesigned and
migrated the platform, restoring the end-to-end data flow.
• Vendor never supplied a data model or interface spec: derived the telemetry payload contract from their device description,
built a Raspberry Pi emulator of their device, and flagged the missing test data in every status report.
• Owned frontend, backend, CI/CD and feature design; started the project from scratch and onboarded and mentored
additional engineers (React Native, QA).
Tech: .NET Core 6, WebAPI, Azure IoT Hub, React, React Native, Azure CI/CD, SCSS, MobX
Harward Media — Fullstack Engineer | 12/2020 – 03/2021
• Built a high-load eCommerce platform for accessories.
• Implemented A/B testing to improve user engagement.
• Developed an admin panel and optimised backend workflows; integrated Google Ads and Facebook Ads APIs.
Tech: .NET Core, EF Core, PostgreSQL, React, React Native, Material-UI, MobX, Azure DevOps
EPAM — Software Engineer | 12/2019 – 08/2020
• Monitoring and control system for oil drilling operations.
• Enhanced UI components and built extensions for the monitoring system.
• Implemented and optimised unit tests for critical system modules.
Tech: WPF, ASP.NET, MS SQL, Oracle Database, Azure DevOps, IoT Hub, LLBL Gen, MEF
Limestone Digital — Frontend Engineer | 01/2019 – 10/2019
• Internal management system for the International Ice Skating Union (ISU).
• Led frontend development with React and TypeScript.
• Integrated SignalR for real-time communication and extended APIs.
• Ran team learning sessions on React, TypeScript and SignalR.
Tech: React, TypeScript, MobX, .NET Core WebAPI, EF Core, SignalR, SASS/LESS
Gamanet a.s. — Software Engineer | 02/2017 – 10/2018
• Security systems for airports, business centres and hotels.
• Developed and tested drivers for IoT-enabled security devices.
• Improved device communication protocols and system reliability.
Tech: .NET, WPF, WF, SVN, Mercurial
Luxoft — Software Engineer | 09/2016 – 01/2017
• Automotive embedded systems for event tracking and navigation; built multi-threaded components and implemented
performance optimisations.
Tech: C++, Qt, Git, multithreading, microcontrollers
Aquarium Co. — Firmware Developer (first commercial role) | 06/2014 – 06/2016
• Developed firmware for Atmel microcontrollers; designed and tested circuit boards for smart home integration.
Tech: Atmel SoC, Raspberry Pi, Fibaro Smart Home
EDUCATION
IT STEP Computer Academy — Software Development | 2016 – 2018
Luhansk National Agrarian University — Master of Veterinary Medicine | 2007 – 2015
LANGUAGES
English — Intermediate (B1)
Ukrainian — Native
Senior Fullstack Engineer | .NET + React | Multi-agent LLM systems
Ukraine (Remote) · [
SUMMARY
Fullstack engineer with 7+ years in .NET and React (12 years in software engineering overall), across IoT, eCommerce,
agriculture, Sitecore and security domains. Most recently designed and shipped to production a memory-backed multi-agent
LLM platform: microservices, gRPC agent orchestration, hybrid RAG and its own CI/CD. Strong in distributed backend
architecture, Azure and modern frontend.
CORE SKILLS
Backend: C#, .NET Core / .NET 10, ASP.NET Core, WebAPI, REST API, EF Core, gRPC, SignalR, Node.js, Express.js,
RabbitMQ, Python (working knowledge)
Frontend: React, React Native, JavaScript, TypeScript, Redux, MobX, Electron, Next.js, SCSS/SASS/LESS, Material-UI
AI / LLM: AI agents and multi-agent orchestration, multi-provider LLM runtime (OpenAI, Anthropic, DeepSeek), RAG
(pgvector, Neo4j, cross-encoder reranking), vector embeddings, prompt versioning, Brier calibration, local model serving, MCP
(Model Context Protocol, working knowledge), AI-assisted development workflows
Architecture: microservices, distributed systems, explicit execution DAGs, SOLID, scalability, code review, mentoring,
Agile/Scrum
Data: PostgreSQL, MS SQL, MySQL, SQLite, Redis, Neo4j
Cloud & DevOps: Azure (primary), AWS (working knowledge), Azure Functions, Azure IoT Hub, Docker, CI/CD (GitHub
Actions, Azure DevOps), YARP, health checks and observability
Testing: xUnit, MS Fakes, Vitest, Playwright, Cypress, unit and integration testing, performance profiling
PROFESSIONAL EXPERIENCE
Independent Fullstack Engineer / Contractor — Self-employed
trading_bot — market research system (own product)
09/2026 – Present · TypeScript, Node.js, Vitest
• Built a research system measuring the link between news and price movement: 94 modules, ~26,900 lines, 1,685
automated tests across 77 files (all green).
• Implemented interval arithmetic (an empty intersection triggers refusal) and Dempster-Shafer evidence theory to separate
conflict from ignorance; refusal is a first-class outcome, not a failure.
• Built a blind-testing harness over historical data: 181 decision steps with the full production logic and no look-ahead — the
system correctly refused all 181 times.
• Enforced a mandatory asOf time parameter across all time-dependent calculations to eliminate look-ahead bias, plus
train/holdout separation and clustered confidence intervals.
• Fixed a 502x performance defect (26,125 ms to 52 ms per call) and rejected a 730x speed-up because a parity lock proved
it leaked future data.
• Served an LLM locally (LM Studio) and calibrated outputs by Brier score; measurement showed no directional skill, so the
model was repurposed as a significance measurer.
veriditarot — multi-agent LLM platform (own product, live in production)
02/2026 – 08/2026 · .NET 10, gRPC, React/Next.js, PostgreSQL, Neo4j, Docker
• Designed and shipped a memory-backed multi-agent platform: 19 .NET microservices, 7 gRPC reasoning agents, a
32-container Docker stack and 4 clients (Next.js web + admin, Expo mobile, Electron desktop).
• Built an explicit artefact-DAG execution engine with bounded concurrency (max 3 concurrent LLM calls per session),
per-agent gates, retries and a circuit breaker keyed by service + method + promptKey; a full Celtic Cross reading fans out to
37 LLM calls in total. Live at veriditarot.com.
• Implemented hybrid RAG: 768-dim pgvector embeddings + Neo4j knowledge graph + LLM-based chunk tagging +
cross-encoder reranking, with a reading-memory feedback loop (durable outbox with reconciliation) that personalises later
sessions.
• Added habit/novelty/decay weighting for cross-session memory ranking, replacing static score constants.
• Shipped CI/CD: a GitHub Actions self-hosted runner builds images to GHCR; the VPS pulls and restarts via docker
compose over a pull-only SSH key. Validated with a 25-concurrent-reading load test: 0 errors, 32% peak CPU on 4 vCPU.
Client contracts — freelance engagements
07/2023 – 12/2025 · .NET Core, React, TypeScript, PostgreSQL
• Multiple short-term fullstack engagements (3-5 months each) for various clients.
• Delivered .NET Core backends and React frontends end to end.
Techery — Fullstack Engineer | 01/2023 – 06/2023
• Fullstack development on client web products: .NET backend services and React frontend.
Tech: .NET Core, React, TypeScript
Mapleton Hill (via EXDST) — Fullstack Engineer | 10/2021 – 12/2022
• Sitecore CMS and public website for Mammoth Mountain Resort (mammothmountain.com).
• Agricultural platform (farm management plus satellite navigation for farming machinery): joined as a fullstack engineer to
bridge the frontend team and the .NET backend.
• Rescued platform performance: reproduced the slowdown by throttling the network, then traced it to 30-40 requests per click
— each of four panels plus the map re-fetched the same deeply nested JSON instead of parsing it once, cascading
re-renders across every panel.
• Fixed the backend half: steady ~30 req/s was served through Azure Functions, which are built for event-driven,
scale-to-zero workloads rather than constant low-latency traffic, and for data that in some cases changes only monthly.
Partially moved off them; roughly halved the request volume.
• Designed client-server architecture for IoT device communication.
• Set up and maintained CI/CD pipelines on Azure.
Tech: .NET Core, Sitecore, React, Redux, Azure Functions, Azure Cloud, CI/CD
Varyence — Fullstack Engineer / Tech Lead | 04/2021 – 09/2021
• Led a smart sports equipment platform end to end: Azure IoT Hub ingests telemetry from the training device, the backend
collects and stores it, and the mobile app shows training data.
• Implemented QR-code pairing between the trainer and the user's phone.
• Found that the vendor-supplied architecture used an Azure IoT tier unfit for production multi-device scale; redesigned and
migrated the platform, restoring the end-to-end data flow.
• Vendor never supplied a data model or interface spec: derived the telemetry payload contract from their device description,
built a Raspberry Pi emulator of their device, and flagged the missing test data in every status report.
• Owned frontend, backend, CI/CD and feature design; started the project from scratch and onboarded and mentored
additional engineers (React Native, QA).
Tech: .NET Core 6, WebAPI, Azure IoT Hub, React, React Native, Azure CI/CD, SCSS, MobX
Harward Media — Fullstack Engineer | 12/2020 – 03/2021
• Built a high-load eCommerce platform for accessories.
• Implemented A/B testing to improve user engagement.
• Developed an admin panel and optimised backend workflows; integrated Google Ads and Facebook Ads APIs.
Tech: .NET Core, EF Core, PostgreSQL, React, React Native, Material-UI, MobX, Azure DevOps
EPAM — Software Engineer | 12/2019 – 08/2020
• Monitoring and control system for oil drilling operations.
• Enhanced UI components and built extensions for the monitoring system.
• Implemented and optimised unit tests for critical system modules.
Tech: WPF, ASP.NET, MS SQL, Oracle Database, Azure DevOps, IoT Hub, LLBL Gen, MEF
Limestone Digital — Frontend Engineer | 01/2019 – 10/2019
• Internal management system for the International Ice Skating Union (ISU).
• Led frontend development with React and TypeScript.
• Integrated SignalR for real-time communication and extended APIs.
• Ran team learning sessions on React, TypeScript and SignalR.
Tech: React, TypeScript, MobX, .NET Core WebAPI, EF Core, SignalR, SASS/LESS
Gamanet a.s. — Software Engineer | 02/2017 – 10/2018
• Security systems for airports, business centres and hotels.
• Developed and tested drivers for IoT-enabled security devices.
• Improved device communication protocols and system reliability.
Tech: .NET, WPF, WF, SVN, Mercurial
Luxoft — Software Engineer | 09/2016 – 01/2017
• Automotive embedded systems for event tracking and navigation; built multi-threaded components and implemented
performance optimisations.
Tech: C++, Qt, Git, multithreading, microcontrollers
Aquarium Co. — Firmware Developer (first commercial role) | 06/2014 – 06/2016
• Developed firmware for Atmel microcontrollers; designed and tested circuit boards for smart home integration.
Tech: Atmel SoC, Raspberry Pi, Fibaro Smart Home
EDUCATION
IT STEP Computer Academy — Software Development | 2016 – 2018
Luhansk National Agrarian University — Master of Veterinary Medicine | 2007 – 2015
LANGUAGES
English — Intermediate (B1)
Ukrainian — Native
Схожі кандидати
-
Fullstack-розробник
Дистанційно, Полтава -
Fullstack-програміст
Дистанційно -
Fullstack-розробник
90000 грн, Дистанційно -
Fullstack-розробник
Дистанційно, Київ -
Fullstack-розробник
30000 грн, Дистанційно -
Fullstack-розробник
Дистанційно, Дніпро , ще 3 міста