EXL MLOps training program

Day 1: Overview of ML

• What is ML? Use cases
• Supervised learning: classification vs regression
• Training and evaluating basic models with scikit-learn

Day 2: Exploratory Data Analysis (EDA)

• Understanding datasets: nulls, outliers, distributions
• Visualization tools (Pandas, Seaborn)
• Feature engineering and transformation

Day 3: Regularization + Deep Learning Overview

• Overfitting, underfitting
• L1/L2 regularization
• Intro to neural networks and deep learning flow

Day 4: NLP & Embeddings

• Bag-of-words vs word embeddings
• TF-IDF, Word2Vec, BERT (overview)
• Hands-on with vectorized NLP classification

Day 5: MLOps + Git, DVC, Feature Store

• MLOps lifecycle: code, data, models
• Git basics for ML
• DVC for data/model versioning
• Feast for feature storage and retrieval

Day 6: Experiment Tracking & Model Registry

• MLFlow/Weights & Biases for tracking experiments
• Logging metrics, parameters
• Model registry and promotion

Day 7: Containerization with Docker

• Dockerfile for ML pipeline
• Packaging training/inference
• Run model container locally

Day 8: Orchestration with Kubernetes

• K8s architecture: pods, deployments
• Deploying and managing model containers
• Minikube or local cluster deployment

Day 9: Monitoring, Logging & Security

• Monitoring performance and drift
• Logging frameworks and best practices
• Securing models and APIs (tokens, rate limits)

Day 10: CI/CD + End-to-End Project

• Building CI/CD workflows
• GitHub Actions for ML
• Deploying a complete ML pipeline
• Showcase of student/real-world projects

---

Bonus Session (Optional): Transformers & LLMs

• Overview of transformers and attention
• LLMs like GPT, BERT, Claude
• Use cases + API demo (OpenAI or Hugging Face)

---