Jupyter Notebook

Hands-on Generative AI: 15 RAG modules from document loaders to guardrails, plus Papeer — a research paper assistant with citations and claim verification. Also LangGraph workflows and LangSmith tracing.

Hands-on course on agentic AI and multi-agent systems - Fundacion AI Granada (July 2026)

This course is designed to take you from the basics to advanced concepts, providing hands-on experience in building, deploying, and optimizing AI models using Langchain and Huggingface. Perfect for AI enthusiasts, developers, and professionals

An AI-powered machine learning system that predicts loan default risk and classifies applicants as safe or risky using multiple ML algorithms.

Explainable, retrieval-based medical question answering chatbot built on NIH-sourced data

GPT-2 train from scratch in pure NumPy, verified against PyTorch. I have written a blog explaining all the math. Check out the website.https://x.com/HarshalsinghCN/status/2055154124535287829?s=20

A method for LLM alignment without RL

End-to-end Employee Attrition Prediction system with data analysis, feature engineering, model comparison, explainable AI (SHAP), FastAPI deployment, and Docker.

Trilha aberta e gratuita, do zero ao avançado, para construir assistentes educacionais inteligentes: NLP, Machine Learning, Deep Learning, Transformers, LLMs, RAG, agentes e Learning Analytics.

Production-grade Agentic AI portfolio exploring autonomous LLM systems, reflection loops, tool usage, multi-agent orchestration, dynamic planning, and containerized AI architectures.

AI-driven Web Search

Claude SDK-powered Spotify playlist generator with iterative refinement and Discogs integration.

Arabic Costs Double on AI — why every Arabic-script language (Darija, MSA, Persian, Urdu) pays a 2x token tax on ChatGPT, Claude & Gemini.

Deep learning framework for automated pneumonia detection from chest X-ray images using transfer learning, data augmentation, and ensemble-based medical image classification. Built with PyTorch using ResNet50, evaluation metrics, ROC analysis, confusion matrices, and visualization for reliable AI-assisted diagnosis.

Real-world AI & Data job market analysis using APIs, ETL, SQL, NLP, and Power BI dashboards.

Personal AI workflows portfolio — Claude-scaffolded analytical systems with built-in verification habits. Early-career economist transitioning toward AI-native finance and analytics.

AI-enabled Investigator Copilot prototype demonstrating alert enrichment, risk explanation, investigator guidance and draft case narrative generation for Financial Crime Operations.

Multi-stage RAG chatbot for mental health support — combining language detection, emotion classification, intent routing, and Qdrant-powered retrieval with Groq LLM streaming responses.

Explainable AI genetic disorder risk screener — Random Forest + GenAI summaries, privacy-first mobile client, offline fallback.

AI-powered Hospital Management Agent using LangGraph, LangChain, Groq LLM, and Python.

A RAG pipeline that reads PDFs and answers questions using LangChain and ChromaDB. The Streamlit app uses Qwen/Qwen2.5-7B-Instruct while a quantized Zephyr-7B LLM is used in the notebook.

Deep learning-based PM2.5 air quality forecasting using LSTM, GRU, and ARIMA to predict 24-hour ahead PM2.5 concentrations in Dhaka.

Agente RAG conversacional sobre IA Generativa con Google Gemini, ChromaDB y LangGraph. Memoria por thread, interfaz Streamlit y base de conocimiento propia. Práctica final del Máster en Data Science e IA.

Exploring Google Gemini API, prompt engineering techniques, system instructions, structured outputs, generation parameters, and chatbot development.

Modeling energy consumption and carbon impact of AI inference workloads. Part of KTH degree thesis

Built a RAG-based recruitment assistant that indexes resume and job description corpora using vector embeddings, enabling semantic candidate search, resume–JD matching, and natural language querying through an LLM-powered conversational interface.

An AI-powered cybersecurity platform designed to detect and prevent phishing attacks using Machine Learning, NLP, URL analysis, screenshot inspection, and browser extension-based real-time protection.

A structured repository documenting my journey of learning Artificial Intelligence and Machine Learning from Python fundamentals to Deep Learning, Generative AI, RAG, AI Agents, MCP, and real-world projects.

A context-aware chatbot built with LangChain, RAG, vector search, and Streamlit for intelligent document based conversations.

Python project for scraping, processing, and analyzing air quality and climate data from the Open-Meteo API using Pandas, NumPy, Git, and PEP 8 best practices.

Agentic AI system for email triage and phishing/scam detection

News Credibility Index (NCI) is a machine learning-based system that evaluates the trustworthiness of news content by combining fake news detection, clickbait analysis, and topic classification using NLP techniques.

A machine learning web application utilizing a fine-tuned MobileNetV3 CNN to classify images of dogs and wolves. Built with TensorFlow and Streamlit for real-time inference. Group CO6 Project.