Taking models from notebook to production and keeping them healthy there.
What has to exist around a model before it can serve real traffic, and the order to build it in.
Watching inputs, outputs and the service — and knowing which alarm means what.
Shadow traffic, canaries, A/B tests and the rollback you test before you need it.
Batching, caching, quantisation, hardware and the arithmetic of a request budget.
Deciding whether a model needs retraining, and building a pipeline that does it safely.
Making a result you can repeat, compare and hand over — with the discipline, not just the tool.
Turning notebooks into steps that run unattended: training, evaluation and promotion as a graph.
Making the environment reproducible: containers, declared infrastructure and environments that match.
Building the evaluation that lets you change a prompt or a model without guessing.
Finding where the money goes in a model-backed system, and the levers that bring it down.
20 in this topic
The practices that take machine learning from notebook to reliable production: versioning, automation, deployment and monitoring.
MLOps & deployment 2 min read 3 Jan 2026 For people
Batch scoring, real-time APIs, streaming and on-device inference: choosing how predictions reach users, and deploying safely.
MLOps & deployment 2 min read 2 Jan 2026 For people
What to monitor after deployment — data, predictions, outcomes and operations — and how to respond when things change.
MLOps & deployment 2 min read 1 Jan 2026 For people
Why every production model needs a version, metadata and lineage, and how a model registry manages promotion and rollback.
MLOps & deployment 2 min read 31 Dec 2025 For people
Continuous integration and delivery adapted to ML: testing code, data and models automatically before deployment.
MLOps & deployment 2 min read 30 Dec 2025 For people
Recording parameters, metrics, data and artefacts for every training run, so results can be compared and reproduced.
MLOps & deployment 2 min read 29 Dec 2025 For people
Reducing latency and cost of inference with batching, caching, quantisation, distillation and the right hardware.
MLOps & deployment 2 min read 28 Dec 2025 For people
Keeping features consistent between training and serving, computing them on time, and avoiding training–serving skew.
MLOps & deployment 2 min read 27 Dec 2025 For people
The operational practices specific to language model applications: prompt versioning, evaluation, observability, cost and safety.
MLOps & deployment 2 min read 26 Dec 2025 For people
When and how to retrain production models — scheduled, triggered or continuous — and how to do it safely.
MLOps & deployment 2 min read 25 Dec 2025 For people
Safely rolling out new models by testing them on live traffic before they fully replace the old ones.
MLOps & deployment 1 min read 24 Dec 2025 For people
Keeping track of which models are running, who owns them, how they were validated and when they need review.
MLOps & deployment 1 min read 23 Dec 2025 For people
Choosing and managing GPUs for training and inference: cloud versus on-premises, sizing, utilisation and costs.
MLOps & deployment 1 min read 22 Dec 2025 For people
The techniques behind fast, affordable LLM inference: batching, KV caching, quantisation and serving frameworks.
MLOps & deployment 1 min read 21 Dec 2025 For people
What to track once an LLM application is live: quality, safety, cost, latency and user feedback.
MLOps & deployment 1 min read 20 Dec 2025 For people
Treating prompts as production artefacts: version control, testing, deployment and rollback.
MLOps & deployment 1 min read 19 Dec 2025 For people
Practical ways to reduce the cost of training, serving and calling models without hurting quality.
MLOps & deployment 1 min read 18 Dec 2025 For people
Beyond accuracy: unit tests, data tests, model behaviour tests and integration tests for ML systems.
MLOps & deployment 1 min read 17 Dec 2025 For people
Running models on phones, devices and sensors: benefits, constraints and optimisation techniques.
MLOps & deployment 1 min read 16 Dec 2025 For people
The shared infrastructure that lets teams build, deploy and monitor models consistently, and how to grow it.
MLOps & deployment 1 min read 15 Dec 2025 For people