Master’s in Model and Data Observability
Online master's programme of 1,900 hours in model and data observability: drift, data contracts, traceability and auditing. 12 modules covering observability foundations, telemetry and instrumentation, data observability, data and concept drift, model performance in production, explainability, LLM and RAG observability, incident management, platform architecture, governance and regulation, and sector cases, with an assigned technical tutor.
Take the controls.
Run a real engineering trade-off from this field. Move the sliders, watch the model react in real time and beat the challenge.
What you will be able to do
Instrument data pipelines and inference services with standard telemetry.
Monitor data quality, freshness and lineage.
Detect data and concept drift in production models.
Measure model performance with and without labels.
Observe LLM- and RAG-based systems.
Manage incidents with SLOs, alerts and postmortems.
Meet the logging and monitoring requirements of AI regulation.
Who it's for
- ML observability engineers.
- MLOps engineers.
- Data reliability engineers.
- Site reliability engineers extending their work to ML systems.
- AI system auditors.
- Data platform managers.
- Engineering or science graduates looking to enter banking and insurance.
What you'll work with
- Regulation (EU) 2024/1689 (AI Act)
- ISO/IEC 42001 (AI management systems)
- NIST AI RMF 1.0 (AI risk management framework)
- ISO/IEC 25012 (data quality model)
- ISO/IEC 5338 (AI system life cycle processes)
- Regulation (EU) 2016/679 (GDPR)
- ISO/IEC 27001 (information security management)
- OpenTelemetry
- Prometheus
- Grafana
- Evidently AI
- OpenLineage
12 modules
1.1Monitoring versus observability1.2Pillars: metrics, logs and traces1.3Data and model life cycle1.4Silent failure modes in ML1.5Cost of data incidents1.6Roles: data, ML and platform engineering1.7Observability maturity1.8Open-source tool landscape1.9SLO-oriented design1.10Case study: postmortem of a degraded model
2.1OpenTelemetry: specification and SDK2.2Instrumenting pipelines2.3Distributed tracing in inference2.4Structured logs and correlation2.5Metrics with Prometheus2.6Cardinality and telemetry cost2.7Trace sampling2.8Business context in telemetry2.9Storage and retention2.10Lab: instrumenting an inference service
3.1Freshness, volume and schema3.2Value distributions and anomalies3.3Lineage with OpenLineage and Marquez3.4Detecting schema changes3.5Pipeline monitoring with dbt3.6Data contracts and expectations3.7Smart alerts with dynamic thresholds3.8Data incident triage3.9Catalogue and owners3.10Practical exercise: quality monitor for a data warehouse
4.1Covariate shift and prior shift4.2Gradual and abrupt concept drift4.3KS, PSI and chi-squared statistical tests4.4Wasserstein and Jensen-Shannon distances4.5Drift in high-dimensional data4.6Drift in text and images using embeddings4.7Reference windows4.8False positives in drift alerts4.9Retraining strategies4.10Lab: drift detection with Evidently
5.1Metrics with delayed labels5.2Performance estimation without labels5.3Segmentation and cohort analysis5.4Probability calibration5.5Fairness in production5.6Latency, throughput and resources5.7Shadow deployment and comparison5.8A/B testing of models5.9Model health dashboards5.10Practical exercise: performance dashboard with Grafana
6.1Global feature importance6.2SHAP and LIME in production6.3Counterfactual explanations6.4Error diagnosis by segment6.5Analysis of out-of-distribution data6.6Uncertainty detection6.7Explanations for audit6.8Computational cost of explanations6.9Communicating with business users6.10Workshop: root cause analysis of errors
7.1Prompt and response traces7.2Relevance and faithfulness metrics7.3Hallucination detection7.4Cost per token and latency7.5Retrieval monitoring7.6User feedback7.7Security: injection and data leakage7.8Continuous evaluation with LLM-as-a-judge7.9Langfuse and Phoenix tools7.10Lab: traceability of a RAG assistant
8.1SLIs, SLOs and error budgets8.2Designing actionable alerts8.3On-call rotas and escalation8.4Runbooks for ML incidents8.5Communication during an incident8.6Blameless postmortems8.7Rolling back models and data8.8MTTD and MTTR metrics8.9Managing recurring problems8.10Case study: data incident drill
9.1The Prometheus, Loki, Tempo and Grafana stack9.2Commercial and SaaS platforms9.3Storing time series and logs at scale9.4Observability on Kubernetes9.5Integration with MLflow and registries9.6Observability at the edge and on devices9.7Multi-tenancy and access control9.8Cost and optimisation9.9Observability as code9.10Practical exercise: deploying an observability stack
10.1Event logging under the AI Act10.2Post-market monitoring10.3Technical documentation and traceability10.4ISO/IEC 42001 and system monitoring10.5Risk management according to the NIST AI RMF10.6Telemetry privacy and GDPR10.7Evidence for auditors10.8Legal retention of records10.9Reports to management10.10Case study: monitoring plan for a high-risk system
11.1Credit scoring and economic drift11.2Industrial predictive maintenance11.3Machine vision in quality inspection11.4Energy demand forecasting11.5Recommender systems11.6Clinical models and medical devices11.7Fraud detection11.8Lessons from public incidents11.9Maturity indicators by sector11.10Workshop: observability strategy for an organisation
12.1Selection of the system to observe12.2Risk map and failure modes12.3SLI and SLO design12.4Instrumentation and telemetry12.5Drift and quality detection12.6Dashboards and alerts12.7Incident management and runbooks12.8Evaluation of the solution12.9Thesis report12.10Master's Thesis: defence before a panel
How you'll learn
The programme is delivered 100% online on the SEIUM campus, with downloadable material, recorded classes and an assigned technical tutor. There are no fixed timetables: you progress at your own pace within each intake, so it fits around your job.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is OpenTelemetry, Prometheus, Grafana, Evidently AI and OpenLineage, in free, academic or evaluation versions. Exercises are based on real procedures: data incident drills, drift detection tests and model monitoring in pre-production environments. Assessment is continuous by module, plus the final Master's Thesis.
Standards covered: Regulation (EU) 2024/1689, ISO/IEC 42001, NIST AI RMF 1.0, ISO/IEC 25012, ISO/IEC 5338, Regulation (EU) 2016/679, ISO/IEC 27001.
Build it end to end
Your final project follows the real engineering workflow, step by step, reviewed by your tutor.
Where this takes you
Requirements
- University degree in engineering, science or architecture, or equivalent demonstrable professional experience.
- Internet connection and a computer capable of running the programme software.
- English sufficient to read technical standards.
Ready to start?
Send your application today. Our admissions team replies within 72 hours with your study plan and funding options.
Frequently asked questions
The programme totals 1,900 hours over 12–14 months. You set your own pace within each intake, and your tutor helps you plan the workload around your job.
You work with OpenTelemetry, Prometheus, Grafana, Evidently AI and OpenLineage, using free, academic or evaluation versions. Your tutor indicates the free alternative wherever one exists.
A SEIUM Institute diploma, awarded on passing the continuous assessment and the Master's Thesis. It is SEIUM's own qualification, not an official state degree; its recognition depends on the programme completed and the regulations of each country.
Yes. Modules passed in a course or postgraduate diploma in the same area are recognised in the higher programme. Ask about the recognition policy when you enrol.
A university degree in engineering, science or architecture (or equivalent professional experience), a suitable computer and English sufficient to read technical standards.
Start dates are set by SEIUM's faculty for each intake. Our admissions team will confirm the next available date when you apply.
Related programmes
Master’s in Reinforcement & Imitation Learning for Control
Online master's programme of 1,900 hours in reinforcement and imitation learning for control: RL/IL, hybrid MPC and safety. 12 modules covering sequential decision-making, tabular and deep…
Master’s in Technical NLP Engineering and Engineering Assistants
Online master's programme of 1,900 hours in technical NLP engineering and engineering assistants: 12 modules covering computational linguistics, Transformer models, RAG over technical documentation, information extraction,…
Master’s in Predictive Maintenance with AI
Online master's programme of 1,900 hours in predictive maintenance with AI: vibration, acoustics, asset health and partial digital twins. 12 modules covering maintenance strategy, reliability, condition…
More in Data, AI, Simulation & Analytics
All Data, AI, Simulation & Analytics programmes → · How to apply · Fees & financing · Scholarships