Hi, I'm Alexander Polyakov. I build AI platforms and LLM systems

USA · dfuture.co · IT, Software Development

• AI Tooling & Developer Experience: Architected and rolled out an internal ecosystem of 10+ AI tools, MCP servers, and extensions for Cursor, Claude Code, and OpenCode. Adopted by 100+ company engineers into daily workflows, cutting feature time-to-market and routine development overhead by ~20%.
• CI/CD & AI QA Automation: Configured automated CI/CD pipelines and automated QA workflows using AI for test scenario generation and staging verification, accelerating release cycles and catching regressions early.
• Production RAG & Vector Search: Built and shipped 5+ production RAG pipelines with hybrid search (dense + sparse / BM25) powered by Qdrant and PostgreSQL (pgvector). Implemented context-aware dynamic chunking and reranking (Cohere Rerank / cross-encoders), reducing retrieval latency and increasing Context Precision.
• Model Context Protocol (MCP) & Tool Calling: Designed the enterprise integration layer via MCP, ensuring secure context retrieval and reliable Function Calling / tool execution across multi-agent systems.
• LLM Evaluation & Guardrails: Built an automated synthetic evaluation pipeline (Ragas, DeepEval, LLM-as-a-Judge). Introduced Faithfulness, Context Recall metrics and semantic guardrails, systematically tracking prompt regressions and decreasing production hallucinations.
• Multi-Agent Orchestration: Designed agentic workflows (Supervisor, Handoff patterns, stateful graphs) to automate complex data analysis and code generation tasks; introduced Spec-Driven Development methodology.
• Architectural Leadership: Authored 20+ Architectural Decision Records (ADRs) on LLM integration and token spend management; led prompt engineering and AI-first engineering workshops for 50+ developers.

• Python (Asyncio, FastAPI)
• Golang
• MCP
• Qdrant
• OpenAI / Claude APIs
• LangGraph
• Ragas
• CI/CD
• Docker
• Kubernetes

B2B & Startup AI Architecture & Engineering · 1 yr 1 mo

• B2B & Startup AI Transformation: Advised founders and CTOs across 6+ tech companies (FinTech, EdTech, LegalTech) on GenAI system architecture, model selection, and LLM integration into existing backends.
• MVP & Production PoC: Shipped 4+ commercial solutions to production: intelligent support agents with Function Calling/Tools, semantic search over private knowledge bases, and automated unstructured document processing pipelines.
• Cost Optimization & Unit Economics: Implemented inference API cost-reduction strategies (semantic caching via Redis/GPTCache, dynamic model routing between lightweight and frontier models, prompt optimization), reducing client token expenditure by 35–50%.
• RAG & Vector Databases: Architected embedding and semantic search pipelines using pgvector and Qdrant; implemented hierarchical chunking and metadata filtering to maximize context retrieval accuracy under strict low-latency SLAs.
• Technical Audit & Reliability: Conducted deep audits of API architectures and message queues under peak loads; built rate-limiting, exponential backoff retries, and graceful fallback strategies for external LLM provider outages.

• Python (FastAPI, Asyncio)
• OpenAI API
• Anthropic Claude
• LangChain
• LlamaIndex
• pgvector
• PostgreSQL
• Qdrant
• Docker
• Redis

Moscow · Absolut Insurance Group · 1 yr 1 mo

• Shipped an end-to-end insurance anti-fraud product from scratch in ~6 months: from requirements gathering to full production deployment.
• Architected and deployed a real-time ML anomaly detection pipeline with model inference via FastAPI, RabbitMQ, and Docker on Kubernetes.
• Led system architecture evolution: migrated MVP monolith into 5+ decoupled critical microservices on Kubernetes.
• Built and led an engineering team of 6: established onboarding processes, Kanban workflows, and cross-team code reviews.

• Python
• FastAPI
• AI / ML
• Keras
• PostgreSQL
• Kubernetes
• Docker
• RabbitMQ
• Ruby on Rails 7

Moscow · FinTech · 1 yr 10 mos

• Engineered a distributed credit application scoring system integrated with 5+ banking APIs under strict latency SLAs.
• Designed high-throughput streaming pipelines and feature extraction services; built a low-latency statistics aggregation microservice in Elixir.
• Optimized complex analytical queries in PostgreSQL and Elasticsearch, resolving bottlenecks in application processing queues.

• Python
• Elixir
• Ruby on Rails
• PostgreSQL
• RabbitMQ
• Elasticsearch
• Kubernetes

Saint Petersburg · Software Development · 1 yr 5 mos

• Developed core modules of an insurance aggregator: resilient billing service with automated payouts and initial algorithmic anti-fraud mechanisms.
• Optimized PostgreSQL transaction loads and configured automated CI/CD pipelines, cutting release delivery cycles by 50%.

• Ruby on Rails
• PostgreSQL
• Redis
• Docker
• CI/CD

Moscow · neuron.digital · Software Development · 1 yr 10 mos

• Built high-performance backend services and ETL pipelines for high-traffic media platforms serving millions of users.
• Implemented an analytical data warehouse and search engines using ClickHouse and Elasticsearch for fast querying and aggregation over large datasets.
• Architected an asynchronous event bus via RabbitMQ for resilient inter-service communication.

• Python
• Golang
• PostgreSQL
• ClickHouse
• Elasticsearch
• RabbitMQ
• Docker

Moscow · polexgroup.ru · 2 yrs 5 mos

• Developed ERP/CRM modules and engineering calculation algorithm services for internal company business processes.
• Designed relational database schemas and optimized complex SQL queries in PostgreSQL to accelerate reporting performance.

• Python
• PostgreSQL
• Linux
• Git