Hands-on Lab

Explore the AI deployment pipeline: Build a simple workflow from data → model → inference → API → application. Identify where training, inference, hardware, APIs, containers, and monitoring fit.

Hands-on Lab

AI Data Processing with Python: Load a dataset, perform NumPy/Pandas operations, clean data, normalize features, visualize basic statistics, and prepare data for ML.

Hands-on Lab

Train Your First Deployable Model: Train a small classification model using Scikit-learn/PyTorch, evaluate it, save the trained model, and perform inference on new samples.

Hands-on Lab

Training vs. Inference: Load a saved model and build a standalone inference script. Compare training and inference time, memory usage, and model inputs/outputs.

Hands-on Lab

Convert a Model for Deployment: Convert a trained PyTorch model to ONNX, verify its outputs, and perform inference using ONNX Runtime.

Hands-on Lab

Optimize an AI Model: Apply techniques such as quantization and/or reduced precision. Compare model size, inference latency, and accuracy before and after optimization.

Hands-on Lab

Run and Use a Small LLM: Load a pre-trained Transformer model, perform text generation/classification, experiment with inference parameters, and measure inference performance.

Hands-on Lab

Build a Simple AI Agent: Create an agent that receives a user query, decides which tool to use, executes the tool, and produces a final response.

Hands-on Lab

Deploy AI on Edge Hardware: Set up a Jetson/edge device, install the required environment, run a pre-trained model, and monitor CPU/GPU/memory utilization.

Hands-on Lab

Real-Time Object Detection: Deploy a YOLO/vision model on an edge device and process a live camera/video stream. Measure FPS and inference latency.

Hands-on Lab

Build an AI API: Use FastAPI to expose a trained model through a REST API. Send an input request and return the model prediction as a JSON response.

Hands-on Lab

Containerize the AI Application: Create a Docker image containing the model and FastAPI application. Run and test the container locally.

Hands-on Lab

Build a Mini MLOps Pipeline: Track experiments/models using MLflow, register a model, deploy a selected version, and monitor basic inference metrics.

Hands-on Lab

End-to-End AI Deployment: Teams build a complete pipeline: data → model → optimization → API → Docker → deployment → inference → monitoring, followed by a demonstration.