AI Specialization Program

• Soft AI Academy
AI Specialization Program
•
Technical Tracks
AI Engineering & Data
•
Week 4

🔹 WEEK 4 — Deep Learning + MLOps + Responsible AI

Theme: Scale, Deploy & Manage Models

Week 4 transitions from:

Classical ML → Neural Networks → Production systems.

Week 4 — 6 Lesson Structure

Lesson 1:

Neural Networks Fundamentals

Focus:

Perceptron intuition

Activation functions

Forward propagation

Backpropagation concept (intuitive)

Why deep learning works

Outcome: Students understand neural network mechanics.

Lesson 2:

Introduction to PyTorch

Focus:

Tensors

Building simple neural network

Loss functions

Optimizers

Basic training loop

Outcome: Students build first neural network in PyTorch.

Lesson 3:

Training Loops & Experimentation

Focus:

Epochs & batching

Learning rate intuition

Debugging unstable training

Logging experiments

Reproducibility mindset

Outcome: Students can manage DL experiments responsibly.

Lesson 4:

Model Versioning & Experiment Tracking

Focus:

Why versioning models matters

Intro to MLflow or DVC

Tracking experiments

Saving models properly

Reproducibility discipline

Outcome: Students understand ML lifecycle management.

Lesson 5:

Model Deployment & Monitoring Basics

Focus:

What is model serving?

API-based model deployment concept

Basic serving with FastAPI (conceptual)

Monitoring performance

Data drift awareness

Outcome: Students understand how models live in production.

Lesson 6: Responsible AI & Bias Detection

Focus:

Bias in datasets

Fairness awareness

Intro to tools (Fairlearn / AIF360 conceptually)

Ethical evaluation

Responsible deployment mindset

Outcome: Students build ethically-aware AI systems.

Week 4 Final Outcome

By the end of Week 4, students can:

✔ Explain neural network fundamentals clearly

✔ Build and train neural networks in PyTorch

✔ Write structured training loops

✔ Manage experiments and track models

✔ Understand model versioning principles

✔ Explain basic model deployment workflow

✔ Identify data drift risks

✔ Detect bias and fairness concerns

✔ Think ethically and production-first

Week 4 produces:

Deep Learning–aware, production-conscious AI engineers.