Skip to the content.

ML Engineering Portfolio

Coursework and projects completed through the Interview Kickstart ML Engineering program — spanning classical ML, deep learning, NLP, transformers, and production LLM systems (RAG, fine-tuning, LoRA/DPO). Each notebook includes detailed reasoning, analysis, and key takeaways — not just code.

Overview

Modules

# Module Key Topics Notebook
1 Build Your First ML Model Linear/Ridge/Lasso regression, feature scaling, cross-validation Open
2 Model Evaluation & Interpretation Precision/Recall/F1, ROC-AUC, confusion matrix, SHAP Open
3 Model Optimization Grid Search, Random Search, Optuna (Bayesian), ensembles Open
4 Neural Networks Basics PyTorch MLP, BatchNorm, Dropout, activation functions Open
5 ML Architectures CNN for images, BiLSTM for text, architecture selection Open
6 Transformer Based Models Self-attention, multi-head attention, transformer blocks Open
7 Deep Dive into LLMs Prompt engineering, LoRA/PEFT, RAG, fine-tuning (SFT/DPO) Open

Projects

Project Score Description
Customer Churn Prediction 92/100 End-to-end churn model: feature engineering → XGBoost → Optuna tuning
Medical Text Classification 88/100 Clinical NLP pipeline: TF-IDF → SVM → BERT fine-tuning
Capstone: RAG Pipeline 95/100 Production RAG: hybrid retrieval, re-ranking, RAGAS evaluation

Completion Summary

How to Run

# Clone the repository
git clone https://github.com/leoniscode/ml-engineering-portfolio.git
cd ml-engineering-portfolio

# Install dependencies
pip install -r requirements.txt

# Launch Jupyter and open any notebook
jupyter notebook

Individual projects have their own requirements.txt for project-specific dependencies:

cd projects/capstone-rag-pipeline
pip install -r requirements.txt

Tech Stack

Core ML: Python, NumPy, pandas, scikit-learn, XGBoost, Optuna

Deep Learning: PyTorch, HuggingFace Transformers, PEFT/LoRA

NLP & LLMs: TF-IDF, BERT, sentence-transformers, LangChain

RAG & Retrieval: FAISS, BM25, cross-encoder re-ranking

Evaluation: SHAP, RAGAS, DeepEval, Evidently AI

Visualization: Matplotlib, Seaborn