Covers the operations side of research data management: FAIR principles, data management plans, metadata standards and persistent identifiers such as DOIs, and dataset versioning in repositories like Zenodo and Dataverse. Also explains reproducible workflows, ethics and consent for research data, HPC containers and large-scale data transfer.
KnowledgeOps applies operational discipline to how teams capture, find and reuse what they know. The course covers Knowledge-Centered Service (KCS), tacit and explicit knowledge, wikis and documentation as code, taxonomy and findability, content ownership and freshness, runbook reuse, knowledge graphs and AI search over internal knowledge.
Covers how operational decisions are designed, automated and improved: decision intelligence, decision modelling with DMN, business rules engines, and combining rules with ML scores. Also covers human-in-the-loop review, decision logging for audit, champion/challenger testing of strategies and measuring decision outcomes.
Covers how analytics and BI are run in production: semantic layers and governed metric definitions, Power BI, Tableau and Looker, and version-controlled analytics code with dbt and LookML. Also explains the dashboard lifecycle and governance, managed self-service, freshness SLAs for reports, usage monitoring and controlling dashboard sprawl.
ModelOps covers the governance and operationalisation of every kind of model an enterprise relies on, from ML and GenAI to rules and optimisation models. It explains model inventories, model risk management under SR 11-7, independent validation and approval, lifecycle states, monitoring and retirement, explainability and fairness reporting, and the basics of the EU AI Act.
Covers how to run LLM applications in production: prompt management and versioning, RAG pipelines and vector databases, and choosing between prompting, RAG and fine-tuning. Also covers offline and LLM-as-a-judge evaluation, guardrails against prompt injection, hallucination mitigation, latency and token cost, and tracing of LLM calls.
Covers the machine learning lifecycle from experiment tracking and reproducible training to deployment and monitoring in production. Explains feature stores, model registries, CI/CD and continuous training pipelines, data and model versioning, serving patterns, drift detection and the Google MLOps maturity levels.
AIOps covers applying big data and machine learning to IT operations, as defined by Gartner: event deduplication and correlation for noise reduction, anomaly detection on metrics and logs, topology-aware root cause analysis, predictive alerting and auto-remediation. It also covers data ingestion from monitoring tools and platforms such as Moogsoft, BigPanda and Dynatrace Davis.
Covers the DataOps Manifesto and how its ideas turn into practice: CI/CD for data pipelines, automated data quality testing with dbt tests and Great Expectations, and data observability across freshness, volume, schema, distribution and lineage. Also explains orchestration with Airflow and Dagster, isolated data environments and data contracts.