Senior AI Engineer

Production AI Agents | RAG, Evals, Memory & Workflows

About Me

I'm an AI Engineer at IntegriAI (EU-based startup), where we built a conversational AI platform that achieved 2× user engagement and 80%+ satisfaction for eCommerce brands.

I specialize in:

  • Production-grade RAG systems with LangGraph
  • Agentic workflows and multi-step reasoning
  • AI systems from PoC to Production, architecture and scaling

Previously, I co-founded and ran Infex Labs — a software development agency for 10 years, delivering 50+ projects for clients who collectively generated $18M+ in revenue and reached 300K+ users.

Project Highlights

View all projects →

Built RAG pipeline enabling document embeddings and semantic search for their chatbot platform.

  • Implemented document chunking strategy for legal documents
  • Built OpenAI embeddings (ada-002) integration
  • Developed pipelines for vector storage and retrieval for semantic search using Pinecone
  • Created FastAPI backend for integration with existing chatbot platform
Impact
  • Enabled querying across legal documents
  • Handling arbitrary document lengths beyond GPT-3's 4K token context limits
PythonFastAPIOpenAI EmbeddingsPinecone

Architected configurable AI chatbot platform enabling deployment of specialized counseling bots (mental health, fitness) with customizable model parameters, system prompts, and tonality via unique bot IDs.

  • Designed multi-tenant architecture with bot ID routing
  • Built dynamic configuration system stored in PostgreSQL
  • Implemented prompt templating and composition system
  • Created parameter management (temperature, model selection, response format)
  • Custom GPT-like functionality built months before OpenAI released Custom GPTs (November 2023)
Impact
  • Rapid bot deployment (days vs weeks)
  • Non-technical stakeholders could configure bots
  • Pre-market innovation
PythonFastAPIOpenAI APIPostgreSQL

Implemented AI-powered document automation system for GSA Schedule application preparation. Automated extraction of information from source documents, alignment with US federal compliance requirements, and generation of 15+ customized application documents.

  • Built multi-phase pipeline: document ingestion and intelligent extraction
  • Implemented compliance requirement mapping for US federal standards
  • Created LLM-powered content generation with appropriate formal language
  • Designed template-based document assembly system
  • Automated generation of 15+ customized application documents
Impact
  • 80% time & cost reduction per application (15+ hours to 3 hours)
  • Enabled 5x client capacity
PythonOpenAI APIDocument ProcessingTemplate GenerationMySQLLaravel