FeedbackShare feedback

← All sections·Learning path

VI

MLOps & Infrastructure17

Advanced

The engineering discipline of running AI in production - reliably, efficiently, and at scale.

Continuous Training

The automated process of regularly retraining ML models on fresh data as part of a production ML system - ensuring that models stay current as the world changes rather than degrading on stale distributions.

Data Drift

The gradual or sudden shift in the statistical properties of data that a deployed ML model receives compared to the data it was trained on - the most common cause of silent model degradation in production.

Data Pipeline

The automated sequence of steps that moves raw data from its sources through transformation, validation, and loading into the storage systems that ML training and inference depend on - the plumbing that makes ML systems run.

All concepts

D

Distributed Training
Training ML models across multiple GPUs or machines simultaneously by partitioning the work - essential for models too large to fit on a single device or datasets too large to process in reasonable time on one machine.

E

Experiment Tracking
The practice of systematically recording every ML training run - logging hyperparameters, code versions, datasets, metrics, and artifacts so experiments are reproducible and comparable, turning trial-and-error into a structured search.

F

Feature Store
A centralised data layer that stores, manages, and serves the computed features that ML models use for both training and inference - eliminating the costly problem of different teams recomputing the same features differently.

G

Gradient Checkpointing
A memory-efficiency technique that trades extra computation for reduced memory usage during training by discarding intermediate activations and recomputing them during the backward pass rather than storing them throughout.

H

Hyperparameter Optimization
The automated search for the best configuration settings for an ML model - learning rate, batch size, architecture choices, regularisation strength - that cannot be learned during training and must be specified in advance.

M

O

ONNX
Open Neural Network Exchange - an open format for representing ML models as portable computation graphs that can be exported from any training framework and run on any compatible runtime, enabling framework-agnostic deployment.

P

Pipeline Orchestration
The automated management of complex multi-step ML workflows - scheduling tasks, resolving dependencies, handling failures, and monitoring execution across the full lifecycle from data ingestion through model deployment.

S

Semantic Cache
A caching layer for LLM applications that recognises when a new query is semantically equivalent to a previous one and returns the cached response rather than re-running inference - cutting cost and latency for applications with repetitive query patterns.

V

Vector Database
A database built to store and search content by meaning rather than exact words - the engine that powers AI search and most retrieval systems.