Vasil Kotsev
Vasil Kotsev

Senior Full-Stack AI Engineer

Open to offers · Member since 16 Nov 2026
Location
Stara Zagora, Bulgaria
Desired salary
Unspecified
Work preference
Remote Only / Full Time
Experience level
Senior

About

Professional summary

I am a Senior Full-Stack AI Engineer with more than eight years of experience building enterprise AI applications, financial SaaS platforms, and intelligent document-processing products. I specialize in designing reliable backend systems, secure APIs, and production-grade integrations that connect AI capabilities to real business operations.

I build backend services primarily with Python and TypeScript, using frameworks such as FastAPI, Django, Node.js, and Fastify. My work includes REST APIs, GraphQL, gRPC, event-driven systems, authentication, authorization, distributed processing, and resilient recovery workflows.

I have extensive experience with LLM and agent engineering, including Anthropic Claude, OpenAI, Gemini, Vertex AI, LangChain, LangGraph, RAG, embeddings, vector search, OCR, and MCP-based enterprise integrations. I focus on controlled tool execution, human approvals, idempotency, validation, and observability to make AI workflows safe and dependable.

I also develop modern React and TypeScript applications for AI-assisted document review, workflow management, and operator consoles. I use real-time communication patterns such as WebSockets and Server-Sent Events to provide users with transparent progress, approvals, and verified outcomes.

My background includes financial document processing, accounting integrations, ERP connectivity, transaction reconciliation, and data extraction workflows. I have integrated platforms such as SAP, NetSuite, Xero, QuickBooks Online, Shopify, Amazon, Stripe, and PayPal.

I am experienced in cloud-native delivery across AWS and GCP, including Docker, Kubernetes, Terraform, Helm, GitHub Actions, and Datadog. I enjoy mentoring engineers, reviewing architecture, and building scalable products that combine practical business value with robust engineering standards.

Skills

29 capabilities

Tech stack & tools

Working toolkit

Experience

Career history

Senior Software Engineer Duvo AI

Architected agent-runtime APIs using TypeScript, Node.js, Fastify, and Python execution tasks. Built resilient workflow persistence with PostgreSQL and Redis, enabling interrupted processes to resume safely without repeating completed business operations.

Orchestrated Anthropic Claude through the Claude Agent SDK and integrated SAP and NetSuite through REST APIs and MCP tools. Implemented authorization controls, human approvals, idempotency, telemetry, and controlled retries; improved interrupted-run recovery from 84% to 92% and reduced production failure diagnosis time from 50 to 29 minutes. Mentored engineers and delivered React operator-console improvements using WebSockets and Server-Sent Events.

Senior Software Engineer Cloud Office

Designed Enterprise AI Agent Platform services with Python, FastAPI, Django, Express.js, and NestJS. Separated authentication, retrieval, integrations, and business operations to support evolving model providers without rewriting customer-facing workflows.

Developed enterprise RAG pipelines using Hugging Face models, LlamaIndex, PostgreSQL, and pgvector, increasing retrieval precision from 72% to 82%. Built intelligent invoice-processing workflows with Google Document AI, OCR, Gemini, Kafka, and gRPC, reducing manual review time from 11 to 8 minutes. Delivered React and Next.js interfaces and deployed services through Docker, Kubernetes, Helm, Terraform, AWS, GCP, and GitHub Actions.

Full Stack Engineer Dext

Extended Dext Prepare document-to-ledger workflows with Ruby on Rails, PostgreSQL, GraphQL, React, and TypeScript. Built review experiences that linked source receipts with extracted supplier, tax, and line-item information for accountant validation before publication.

Refined Python document-intelligence workflows using PyTorch, scikit-learn, and pandas. Improved GraphQL and REST exports to Xero and QuickBooks Online, raising successful exports from 94% to 97% through validation and recovery handling. Normalized commerce data from Shopify, Amazon, Stripe, and PayPal, reducing unmatched payout records from 250 to 190 per month, while supporting AWS delivery practices with Docker, Kubernetes, Terraform, DVC, and Jupyter.

Software Engineer Rossum

Developed Python backend services and REST API resources for a document-processing platform, supporting organizations, workspaces, queues, schemas, and annotations. Maintained clear boundaries between document lifecycle workflows and AI extraction services.

Built customer-specific extension workflows with Python and Node.js, webhooks, and ERP/RPA connectors, reducing integration setup time from 8 to 6 days. Stabilized asynchronous processing with RabbitMQ, PostgreSQL, Redis, MongoDB, AWS, and Kubernetes, reducing downstream delivery failures from 2.5% to 1.9%. Migrated integrations to a backward-compatible Document Management API supporting JSON, CSV, and XML exports.

Education

Learning history

Technical University of Sofia

Bachelor of Science, Computer Science

Bachelor of Science in Computer Science.

This professional hasn’t added portfolio projects yet.

This professional hasn’t listed any services yet.

Jobs Talent AI Tools Salaries
Menu