# Ram Pakanayev - AI Engineer
A professional portfolio and technical roadmap for AI Engineering in 2026.

## About
Ram Pakanayev is an AI Engineer specializing in production agentic AI systems, Model Context Protocol (MCP) tooling, multi-model LLM orchestration, and LLM evaluation. Based in Tel Aviv, Israel. Currently an AI Engineer at Elementor, contributing to Angie, its MCP-based agentic AI assistant embedded in the WordPress editor (80,000+ active installations).

## Key Expertise
- Agentic AI & MCP Tooling (tool & server design, self-validation, evaluation harnesses)
- LLM Evaluation (LLM-as-judge across quality dimensions, Langfuse tracing)
- Multi-Model LLM Orchestration (routing and A/B-testing across Claude, Gemini, GPT via Statsig)
- GraphRAG Architecture (Neo4j + pgvector hybrid retrieval over code)
- Multi-Agent Orchestration (LangGraph plan/generate/review workflows)
- Hebrew Voice AI (real-time STT/TTS with SageMaker containerized models)
- AI Observability (OpenTelemetry, Langfuse, Statsig, Sentry, Mixpanel)

## Primary Resources
- [AI Engineer Roadmap 2026](https://rampakanayev.com/ai-engineer-roadmap): The definitive guide to modern AI engineering covering GraphRAG, Multi-Agent Systems, and AWS SageMaker BYOC.
- [LLM Token Cost Calculator](https://rampakanayev.com/llm-cost-calculator): Free tool to compare API pricing across GPT-4, Claude, Gemini, AWS Bedrock, and Azure.
- [Technical Blog](https://rampakanayev.com/blog): Deep-dives on LangGraph, GraphRAG, and MLOps.
- [Portfolio](https://rampakanayev.com/): Professional portfolio with production AI projects.
- [Enterprise Code Intelligence Platform](https://rampakanayev.com/#projects): Secure B2B GenAI platform with dual GraphRAG + Vector architecture.
- [Consensus Chat](https://rampakanayev.com/#projects): Multi-Agent debate platform orchestrating Claude, GPT-4, Gemini, and Grok.
- [Hebrew Voice Agent](https://rampakanayev.com/#projects): Real-time voice-to-voice AI with SageMaker BYOC deployment.

## Technical Articles
- [Human-in-the-Loop in LangGraph](https://rampakanayev.com/blog/langgraph-human-in-the-loop): Implementing approval gates, breakpoints, and feedback loops in multi-agent workflows.
- [Serverless to Self-Hosted Migration](https://rampakanayev.com/blog/neo4j-vs-pgvector-graphrag): How to substantially reduce AWS costs by migrating from Neptune/OpenSearch to Neo4j/Pgvector on EC2.
- [SageMaker Hidden Costs](https://rampakanayev.com/blog/sagemaker-provisioned-hidden-costs): Avoiding billing surprises with provisioned inference endpoints.

## Technologies
MCP, LangGraph, LLM-as-Judge, Langfuse, Statsig, Neo4j, pgvector, AWS Bedrock, AWS SageMaker, Claude, GPT-4, Gemini, FastAPI, NestJS, Next.js, Python, TypeScript, PHP, Docker

## Availability
Ram is open to new opportunities: full-time AI engineering roles, remote collaboration with international and US-based teams, and freelance or consulting engagements. The best way to start a conversation is by email at rampakanayev@gmail.com or on LinkedIn.

## Contact
- Website: https://rampakanayev.com
- LinkedIn: https://linkedin.com/in/ram-pakanayev
- GitHub: https://github.com/RamPakanayev
- Email: rampakanayev@gmail.com