Skip to content
View StamKavid's full-sized avatar

Block or report StamKavid

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
StamKavid/README.md

Portfolio Substack

Hey, I'm Stamatis 👋

AI Innovation Lead at Pfizer, where I run a team of AI engineers focused on one specific problem: getting AI from "cool demo" to actually working in production. We scout emerging tech, stress-test it against enterprise reality, and figure out what's ready for prime time — and what isn't.

I founded The Last AI Mile, a technical community for AI engineers who are tired of the gap between proof-of-concept excitement and production deployment pain. If you've ever built something brilliant in a notebook that died the moment it hit a real pipeline... yeah, that's what we talk about.


📂 Projects

Three specialized AI agents — Financial Analyst, Research Analyst, and Investment Advisor — collaborate to analyze crypto markets end-to-end. Built with CrewAI, it uses tree-based feature selection across 85+ technical indicators and time-series prediction to generate comprehensive investment reports. MBA Dissertation project.

Python CrewAI yfinance scikit-learn Pandas Seaborn

An 8-chapter structured course for mastering prompt engineering with Claude AI. Covers prompt architecture, context engineering, few-shot learning, output formatting, and production optimization. Includes working code examples, interactive exercises, mermaid diagrams, and an assessment engine.

Python Claude API Prompt Engineering

10-chapter deep dive into Large Language Models following Andrej Karpathy's methodology. Covers the complete pipeline from data processing and tokenization through transformer architecture, pre-training, supervised fine-tuning, RLHF, and deployment — with hands-on code and visual diagrams throughout.

Python LLMs Transformers RLHF

Full stack healthcare app correlating patient interactions with hospitalization duration using the MIMIC-III clinical database. Flask REST API backend, React frontend, and ML prediction pipeline — data science through deployment in a single repo.

Python Flask React scikit-learn CatBoost XGBoost


🛠️ Tech Stack

AI/ML & GenAI — LLM orchestration & agent frameworks · RAG pipelines · Document AI & parsing evaluation · TensorFlow · PyTorch · Keras · scikit-learn · CrewAI · Model evaluation & benchmarking

Languages & Data — Python · SQL (Snowflake, Redshift, Athena) · R · Pandas · NumPy · SciPy

Web & APIs — React · Dash · Streamlit · FastAPI · NodeJS

Infra & Tools — AWS · Dataiku DSS · Git · GitHub (CODEOWNERS, PR templates, code review workflows) · Jira


🌐 Find Me

LinkedIn Substack


If you're building AI that needs to survive contact with production, let's talk.

Pinned Loading

  1. FinAgent FinAgent Public

    AI multi-agent system for comprehensive Bitcoin (BTC) analysis, combining financial news, market performance, and AI-driven price predictions for investment recommendations.

    Jupyter Notebook 52 13

  2. karpathy-llm-deep-dive karpathy-llm-deep-dive Public

    Complete deep dive into Large Language Models - from tokenization to deployment. Following Andrej Karpathy's methodology with hands-on examples and practical implementations.

    10 1