Evgenii Akhrimenko
AI Engineer
4+ years exp
4+ years ML exp
Node JS
JavaScript
Python
TypeScript
System Architecture: 5/10
API Design: 4/10
Reliability & Observability: 4/10
Active 18 days ago
+7 (985) 8329828 Invite to interview
Message
Download CVCV
Overview
Technical skills
Timeline
Roles
Overview
A backend API-focused engineer at a Senior (tier 4.0) level who builds LLM-backed retrieval services and realtime voice integrations. The strongest proven skill is designing a hexagonal API architecture with concrete adapters and test coverage, evidenced by app/api/dependencies.py, app/adapters/vector_qdrant.py and the tests under tests/. Public code does not show large-scale distributed operations, formal DB migration histories, or full production infra (kubernetes/CI/CD/monitoring) for evaluation.
Technical skills
Node JS
JavaScript
Python• Senior • 4y+
TypeScript• Senior • 4y+
Python
FastAPI
Mypy
HTTPX
Pydantic
Databases
PostgreSQL• 4y+
Qdrant• 4y+
DevOps
CI/CD
Git
WebRTC
Docker• 4y+
Rest API• 4y+
Vercel• 4y+
AI/ML
Llama
Mistral
LangChain• 4y+
LLM• 4y+
Prompt Engineering• 4y+
RAG• 4y+
Function Calling• 3y+
Frontend
Lighthouse
PostCSS
Next.js• 4y+
React.js• 4y+
Tailwind CSS• 4y+
Timeline
Founder & AI Automation Engineer
•
Executive
Ahrim AI Lab
•
Cofounder
Founder of Ahrim AI Lab providing end-to-end AI automation and solutions for international clients. Build workflow automations that integrate CRMs, messengers and external APIs with monitoring and error handling. Develop Python data scrapers/parsers and full-stack web apps using Next.js/React, REST APIs and PostgreSQL, deploying with Docker and Vercel.
LLM
RAG
LangChain
Qdrant
Function Calling
Prompt Engineering
Python
Next.js
React.js
TypeScript
Tailwind CSS
PostgreSQL
Rest API
HubSpot
Docker
Vercel
Brokerdesk-ai
Desktop app
Full-stack AI-платформа для брокерского бизнеса, собранная end-to-end на Next.js / TypeScript. Голосовой агент работает прямо в браузере через OpenAI Realtime API по WebRTC — с обработкой перебиваний, function calling и извлечением структурированного лида из диалога; рядом чат, квалификация лидов и интеграции с CRM. Ключевую безопасность решил серверно: ephemeral-токены выпускаются на бэкенде, основной ключ не уходит в браузер. Docker + CI/CD, деплой под ключ. Продукт, фронт, бэкенд и инфра — один инженер, а не три.
Senior Backend Developer
Confidence: High API Engineer
A backend API-focused engineer at a Senior (tier 4.0) level who builds LLM-backed retrieval services and realtime voice integrations. The strongest proven skill is designing a hexagonal API architecture with concrete adapters and test coverage, evidenced by app/api/dependencies.py, app/adapters/vector_qdrant.py and the tests under tests/. Public code does not show large-scale distributed operations, formal DB migration histories, or full production infra (kubernetes/CI/CD/monitoring) for evaluation.
API Design
4/10
How well APIs are designed
API design is pragmatic with clear schemas, dependency-injected services and tests for endpoints, but lacks explicit versioning, advanced idempotency patterns or a documented error contract beyond HTTP status codes.
Evidence
rag-system/app/api/routes/ask.py
rag-system/app/api/schemas.py
rag-system/tests/test_api.py
Data Layer & Database
3/10
Working with databases
Data layer shows deliberate adapter patterns for vector storage and an in-process vector store plus tests; however there is no migration history, no explicit transaction/isolation handling and no complex tuning of indexes/queries.
Evidence
rag-system/app/adapters/vector_qdrant.py
rag-system/app/adapters/in_memory.py
rag-system/tests/test_adapter_qdrant.py
Scalability & Performance
3/10
Handling load and speed
Some scalability and resilience concerns are addressed (ephemeral token minting, per-IP rate limiting, LLM fallback and timeouts), but there is no queue-based decoupling, cache-invalidation strategy, or measured load-testing evidence.
Evidence
ai-portfolio/code-samples/realtime-token-route.ts
ai-portfolio/code-samples/llm_client.py
rag-system/app/adapters/vector_qdrant.py
System Architecture
5/10
Overall system structure
Clear hexagonal architecture with domain ports, adapters and services and reasoned DI wiring - a deliberate modular decomposition suitable for swapping LLMs/vector stores; this demonstrates thoughtful system architecture for API-first services.
Evidence
rag-system/README.md
rag-system/app/api/dependencies.py
rag-system/app/domain (ports) + app/adapters + app/services structure
Security & Auth
3/10
Protecting data and access
Good security hygiene in spots - ephemeral tokens, avoiding client-side main keys, and guarded secret placeholders - but full auth/token lifecycle, revocation, or a dependency audit are not present in the public surface.
Evidence
ai-portfolio/code-samples/realtime-token-route.ts
ai-portfolio/code-samples/voximplant-voice-agent.js
ai-portfolio/code-samples/llm_client.py
Reliability & Observability
4/10
Stability and monitoring
Reliability is well-considered: adapters wrap remote errors into domain exceptions, unit tests exercise failure modes, the voice agent handles graceful finalization, and LLM client retries fall back - but there is no full observability stack or alerting shown.
Evidence
rag-system/app/adapters/vector_qdrant.py
ai-portfolio/code-samples/voximplant-voice-agent.js
ai-portfolio/code-samples/llm_client.py
Expertise
Backend AI & LLM• Senior
Databases & Vector Storage• Middle
Messaging & Real-time• Senior
Microservices & API Architecture• Middle
Python• Middle
Industries
Financial Services• Middle
Real Estate• Middle
Technologies
Python• Senior • 4y+
FastAPI
Pydantic
HTTPX
Qdrant• mentioned only
Stack• mentioned only
Recommendations
- Lead development of RAG and LLM adapter work - implement new adapters and maintain the hexagonal boundaries.
- Build realtime voice integrations and ephemeral-token flows for browser/agent clients, extending the existing token-route and VoxEngine patterns.
- Own productionization of vector-search services - add monitoring, access control, migration/versioning and capacity planning for the vector store.
- Implement end-to-end observability (traces/metrics/alerts) and documented error contracts for API consumers.
Repositories
The developer's experience in this domain has been verified based on AI analysis of the following repositories: