💼 All Projects
24 hands-on projects spanning all 9 phases — from environment setup to a deployed capstone AI app. Each project has a reference doc, resources, and level rating.
Phase 0 · Foundation
Dev Environment Setup
Set up Python, notebooks, and Git so every later AI project is reproducible and collaboration-ready.
AI Landscape Mind Map
Research core AI domains and build a visual map linking problems, models, datasets, and real-world use cases.
Phase 1 · Python + DSA
Titanic EDA Analysis
Do structured EDA, feature engineering, and baseline classification on the Titanic dataset with clear insights.
LeetCode 75 Solutions
Solve curated DSA problems and document pattern-based approaches with clean complexity explanations.
CLI Task Manager
Build a production-style command-line app using argparse, JSON persistence, and modular Python code.
Phase 2 · Math + Stats
Gradient Descent Visualizer
Implement gradient descent from scratch and visualize how learning rate impacts convergence on loss surfaces.
PCA on MNIST
Reduce MNIST dimensionality with PCA and quantify the tradeoff between compression and predictive performance.
Bayesian A/B Tester
Build a Bayesian A/B testing engine that estimates probability of lift and decision confidence.
Phase 3 · Core ML
House Price Prediction
Train a robust tabular regression pipeline with feature engineering and cross-validated model tuning.
Customer Churn Classifier
Predict customer churn risk and explain model behavior using SHAP for actionable retention insights.
Movie Recommender
Train collaborative filtering recommenders and evaluate ranking quality on user-item interactions.
Phase 4 · Deep Learning
CNN Image Classifier
Build and tune a CNN image classifier with data augmentation, checkpoints, and class-wise error analysis.
LSTM Sentiment Analysis
Train LSTM-based text classification and compare sequence models against simpler NLP baselines.
Mini-GPT from Scratch
Implement a compact transformer language model to deeply understand tokenization, attention, and training loops.
Phase 5 · Generative AI
RAG Document Chatbot
Build retrieval-augmented Q&A on custom documents with source-grounded responses and failure handling.
Fine-tune Llama 3 (QLoRA)
Adapt an open LLM using parameter-efficient fine-tuning with low-VRAM quantized training.
Semantic Search Engine
Create embedding-based search with vector indexing, relevance tuning, and retrieval quality metrics.
Phase 6 · Agentic AI
Web Research Agent
Orchestrate planning, search, and synthesis nodes to return concise source-backed research briefs.
Multi-Agent Code Reviewer
Coordinate specialist agents to review pull requests for bugs, style, testing gaps, and security risks.
Customer Support Agent
Design a support assistant with memory, policy grounding, and reliable escalation triggers.
Phase 7 · System Design
ML REST API (Deployed)
Deploy a model-serving API with validation, containerization, and operational guardrails.
LLM App + Monitoring
Ship an LLM app with monitoring for traces, latency, costs, and output quality drift.
Phase 8 · Portfolio
Capstone AI App
Build a polished end-to-end AI product with clear user value, architecture clarity, and demo readiness.
Kaggle Competition Entry
Execute disciplined competition experiments and present a leaderboard-improving workflow with reproducibility.