Senior Full-Stack Developer with deep experience building scalable enterprise systems, now focused on GenAI innovation. Proven track record in full-stack development (React, Node, Java, Python), monorepo architecture, and rapid prototyping.

Check out enterprise-grade cloud sandboxes for AI agents at Leap0.dev. Instead of running model-generated code on trusted infrastructure, teams run it in isolated Firecracker microVM sandboxes with APIs for commands, files, git, terminals, and telemetry. This gives developers real automation power while giving enterprises the controls they need: network boundaries, org-level governance, audit-ready evidence, and predictable operations. The result is a practical path from AI demos to production-grade autonomous workflows, without exposing core systems to untrusted runtime behavior.

Finished AI Newline AI Accelerator Program covered Python, statistics, prompt engineering, RAG systems, transformer architectures, multimodal apps, fine tuning, model evaluation and agentic AI.

Some of my AI projects:

• Basic RAG with Eval RAG pipeline using Python, Ollama, and Vector DB Pinecone for ingestion, embedding, and retrieval and metric calculation.
• Dating Compatibility Embedding - fine tune domain-specific model using contrastive loss on synthetic data with scikit-learn, SentenceTransformers, PyTorch.
• LLM-as-Judge for evaluationing Data QA for DIY/Repair. AI chat with Vercel AI Toolkit, ShadCN UI, FastAPI, OpenAI, Pydantic.
• Tennis Email Multi-agent CrewAI for automating tennis league match organization using custom and LangChain Gmail tools, and GCP.
• Digital Clone Agent Multi-agent LangGraph agent that learns your communication style, retrieves info from documents, evaluates responses for style and grounding, and books meetings if confidence is low.
• Job Trucking Job trucking tool using RAG with Ollama, LangChain, ChromaDB, FastAPI, and Streamlit. Airflow for RAG automation.
• Recipe Keeper A smart recipe management and search application powered by RAG. It uses a hybrid search approach combining structured graph data with Neo4j and vector embeddings with Weaviate to allow users to find recipes by ingredients, nutritional info, or even by image similarity.
• Building an AI-driven Contract Agent for a fireproofing and spray foam company using LangGraph and LlamaParse (LlamaIndex) to detect and parse new project contracts, enable chatting with Streamlit, and automate pre-production task management with Python and FastAPI on AWS.
• Medical receipts processing with Zapier AI agents, integrating Gmail, WhatsApp, Google Sheets, image/PDF parsing.
• Coach system handling a large number of concurrent curls via Nginx, and built with Python, Shell, and Telegram bot.