I am an AI Developer specializing in LLM-based solutions, Generative AI frameworks, and business process automation. With a background in Computer Engineering from Politecnico di Milano and 7+ years of experience in IT systems development and management, I bridge the gap between cutting-edge AI innovations and production-ready enterprise tools.
- LLM Solutions & GenAI Frameworks: Researching and implementing Python-based AI architectures using Cheshire Cat and Open WebUI.
- LLMOps & Observability: Routing model traffic through a LiteLLM gateway for unified multi-provider access, and instrumenting it with Langfuse for tracing, cost tracking, and output-quality evaluation.
- Document Intelligence & Prototyping: Developing custom AI tools to streamline enterprise document management and internal processes.
- AI Innovation & Strategy: Experimenting with emerging AI technologies to evaluate their integration and strategic value within business contexts.
- AI-Assisted Development: Building with coding agents like Claude Code and OpenCode, across multiple LLM providers.
- Programming Languages:
Python|SQL - ML & Data Science:
TensorFlow|Scikit-Learn|Pandas|NumPy - RAG & AI Agents:
LangChain|LangGraph|LlamaIndex|CrewAI - GenAI Platforms & Interfaces:
Cheshire Cat|Open WebUI|AnythingLLM - LLMOps, Automation & DevOps:
LiteLLM|Langfuse|n8n|Docker|Git - AI Coding Assistants:
Claude Code|OpenCode - LLM Providers:
Claude|OpenAI|Gemini|DeepSeek|Groq - Enterprise Systems:
Magento|Shopify|PIM
- M.Sc. in Computer Science & Engineering β Politecnico di Milano (2015 - 2017)
- B.Sc. in Computer Engineering β Politecnico di Milano (2011 - 2014)
- repomind: Python multi-agent assistant that indexes source code and answers questions about it, with full LLM tracing.
- chopthesong: Streamlit app that breaks a song into its parts (genres, lyrics summary, samples, Spotify links), driven by a LangGraph pipeline on Groq.
- nba-weekres-bot: Weekly NBA recap bot on Telegram.
- met-me: LLM-powered Telegram bot with RAG and emotion detection.
University projects (Politecnico di Milano)
- PiCoNet: Built a neural network in Python & TensorFlow to compute the pitch of human voice signals (NLP).
- Recommender System Challenge (Kaggle): Job recommendation algorithm based on user search history in Python.
- Taxi Trajectory Prediction (ECML/PKDD): Destination prediction using data mining techniques in R.
- LinkedIn: linkedin.com/in/mttpzz
- Email: pozzi.teo@gmail.com
