Lectures
Each lecture lists its slides and cheat sheet, then its recording, then practice: the lab that followed it, notebooks, demos and code. Slides and labs come from stt-ai-teaching.
Data engineering
Lectures 1–6| Lecture 01 | Data collectionLab: data collection. |
|
| Lecture 02 | Data collection: HTML parsing |
|
| Lecture 03 | Data validationLab: data validation. |
|
| Lecture 04 | Data labellingLab: data labelling, with a challenge to build a mini Label Studio. |
|
| Lecture 05 | Optimising data labelling: active learning, weak supervision and LLMsLab: active learning and weak supervision. |
|
| Lecture 06 | Data augmentationLab: an introduction to data augmentation. |
Model development
Lectures 7–10| Lecture 07 | LLM APIs and multimodal AILab: data augmentation and LLM APIs. |
|
| Lecture 08 | Model evaluation: cross-validation, bias–varianceLab: cross-validation with scikit-learn and from scratch. |
|
| Lecture 09 | Tuning, AutoML and experiment trackingLab: hyperparameter tuning with grid search, random search and Optuna. |
|
| Lecture 10 | Reproducibility: Docker, seeds and trackingLab: Docker, seeds and TrackIO. |
Production
Lectures 11–13| Lecture 11 | Data drift and model monitoringLab: a data drift demo, and a Streamlit deployment demo for Assignment 4. |
|
| Lecture 12 | Profiling and quantizationLab: profiling and quantization. |
|
| Lecture 13 | Building AI agents from scratchLab: AI agents, including an advanced NumPy-aware calculator exercise; a second session continued it and wrapped up the course. |
|
Course summary |
Covered in labs
3 topicsFrom script to web app |
||
CI/CD and automation |
||
APIs and demos |
Cheat sheets are from nipunbatra.github.io. Recordings are available for lectures 1–7.