Master’s in Federated Learning & Privacy-Preserving ML Engineering
Online master's programme of 1,900 hours in federated learning and privacy-preserving machine learning engineering. 12 modules covering ML privacy threats, federated learning foundations, heterogeneity, communication and scalability, vertical and cross-organisation federation, differential privacy, secure aggregation, attacks and defences, foundation models, federated MLOps and regulation, with an assigned technical tutor.
Master’s in Federated Learning & Privacy-Preserving ML Engineering — studying from Ireland
Ireland is one of Europe's leading locations for data centres, cloud services and data-driven businesses, with major technology companies operating large engineering and analytics teams in Dublin and beyond. Research centres including the Insight Research Ireland Centre for Data Analytics and the ADAPT Centre advance work in machine learning, data science and simulation. Engineers in this area build data pipelines, simulation models, predictive analytics and AI systems for industry and public services. Compliance with European data protection and AI regulation, overseen by national regulators, is an important part of the professional landscape.
Fees for Ireland: €5,920 (Euro), with instalments and employer sponsorship available.
Relevant bodies in Ireland
Typical roles
Questions from Ireland
Which Irish bodies relate to recognition of data and AI engineering qualifications?
Ireland has no statutory licence for data or AI engineers. Engineers Ireland covers engineering professional standards, and the Data Protection Commission regulates personal data handling under GDPR, which is relevant to data work. SEIUM's qualifications are private and not accredited by any national body, so their recognition depends on employers or the relevant body, which you should check beforehand.
Do Irish tech employers value a private online AI or analytics qualification?
Ireland hosts many large technology, data and analytics employers, and they generally prioritise demonstrable skills, portfolios and practical projects. A private qualification may support your profile, but how it is weighed is up to each employer, and no particular outcome can be guaranteed. Build a visible body of work alongside any study.
Is the course manageable alongside work, and what language and time commitment is involved?
The programme is fully online, taught in English and requires a strong command of technical English. You can study from Ireland without a visa. Master's programmes are 1,900 hours, postgraduate diplomas 800 and specialist courses 300. An assigned technical tutor supports you, fees can be paid in instalments, and employer sponsorship may be possible, which is worth asking your employer about.
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
Design cross-device and cross-silo federated learning systems.
Address data and system heterogeneity in federated settings.
Apply differential privacy to model training.
Implement secure aggregation and cryptographic techniques.
Evaluate attacks and defences in federated learning.
Deploy federated solutions in production.
Analyse the regulatory fit of PPML projects.
Who it's for
- Federated learning engineers scaling training across devices and silos.
- ML privacy engineers applying differential privacy and secure aggregation.
- PPML researchers working on attacks, defences and auditing.
- Secure AI systems architects designing production federated platforms.
- Healthcare data scientists collaborating across hospitals without sharing data.
- Privacy technology consultants advising on PETs and regulation.
- Engineering or science graduates aiming to enter the healthcare sector.
What you'll work with
- Regulation (EU) 2016/679 (GDPR)
- ISO/IEC 27701 (privacy information management)
- Regulation (EU) 2022/868 (Data Governance Act)
- ISO/IEC 29100 (privacy framework)
- ISO/IEC 20889 (privacy-enhancing data de-identification techniques)
- Regulation (EU) 2024/1689 (AI Act)
- IEEE 3652.1 (architectural framework for federated machine learning)
- Flower
- NVIDIA FLARE
- Opacus
- OpenFHE
- TensorFlow Federated
12 modules · 1900 h
1.1The ML life cycle and exposure points1.2Personal data in training and inference1.3Threat models1.4Honest-but-curious and malicious attackers1.5Privacy-enhancing technologies1.6Trade-offs between privacy, utility and cost1.7Use cases that motivate federated learning1.8State of the art in PPML1.9Regulatory privacy requirements1.10Case study: threats in a medical diagnosis system
2.1Formulation of federated optimisation2.2FedAvg and FedSGD2.3Cross-device versus cross-silo federated learning2.4Horizontal, vertical and transfer federated learning2.5Client selection and sampling2.6Rounds, local epochs and convergence2.7Federated evaluation metrics2.8Simulation of federated scenarios2.9Framework ecosystem2.10Lab: first federated training with Flower
3.1Non-IID data and its types3.2FedProx and proximal regularisation3.3SCAFFOLD and drift correction3.4Normalisation in federated settings3.5System heterogeneity3.6Per-client federated personalisation3.7Federated learning with client clustering3.8Fairness across clients3.9LEAF and FLamby benchmarks3.10Practical exercise: comparison of FedAvg, FedProx and SCAFFOLD
4.1Communication cost4.2Compression and quantisation of updates4.3Gradient sparsification4.4Asynchronous federated learning4.5Hierarchical and decentralised topologies4.6Peer-to-peer learning4.7Federated learning at the edge and on mobile devices4.8Dropout tolerance4.9Large-scale deployments4.10Workshop: update compression in Flower
5.1Feature partitioning5.2Private entity alignment5.3Split models and split learning5.4Federated decision trees5.5Vertical federated learning with encryption5.6Consortia and governance5.7Incentives and contribution valuation5.8Federated learning in banking and insurance5.9Federated learning in healthcare across hospitals5.10Lab: vertical federated logistic regression
6.1Definition and properties6.2Additive noise mechanisms6.3Rényi privacy accounting6.4DP-SGD and gradient clipping6.5User-level differential privacy6.6Differential privacy in federated learning6.7Private hyperparameter tuning6.8Impact on accuracy and fairness6.9Using the Opacus library6.10Practical exercise: DP-SGD training in Opacus
7.1Secure aggregation protocols7.2Shamir's secret sharing7.3Homomorphic encryption in aggregation7.4Multi-party computation for training7.5Trusted execution environments7.6Private inference as a service7.7Cost and latency7.8OpenFHE and CrypTen libraries7.9Design of hybrid protocols7.10Lab: secure aggregation with secret sharing
8.1Gradient-based data reconstruction8.2Membership inference8.3Property inference8.4Data and model poisoning8.5Backdoor attacks8.6Robust aggregation: Krum and median8.7Detection of malicious clients8.8Privacy auditing8.9Evaluation of defences8.10Workshop: gradient reconstruction attack and defence
9.1Federated fine-tuning of large models9.2LoRA and adapters in federated learning9.3Federated language models9.4Privacy in LLM fine-tuning9.5Memorisation and extraction9.6Synthetic data as an alternative9.7Machine unlearning in federated learning9.8Cost on devices9.9Research trends9.10Practical exercise: federated fine-tuning with LoRA
10.1Production architecture10.2Client orchestration10.3Model version management10.4Monitoring without data access10.5Federated evaluation in production10.6Infrastructure security10.7Deployment with NVIDIA FLARE10.8Certificates and identities10.9Auditing and traceability10.10Lab: cross-silo federated deployment with NVIDIA FLARE
11.1GDPR and federated learning11.2Legal bases and responsibilities11.3Data protection impact assessment11.4Anonymisation under the GDPR11.5Data Governance Act11.6European Health Data Space11.7The AI Act and training data11.8Consortium agreements11.9ISO/IEC privacy standards11.10Case study: governance agreement for a federated consortium
12.1Definition of the federated scenario12.2Threat model12.3Design of the training protocol12.4Privacy and security mechanisms12.5Implementation and simulation12.6Evaluation of utility, privacy and cost12.7Regulatory analysis of the scenario12.8Conclusions and future work12.9Master's Thesis report12.10Master's Thesis: defence before a panel
How you'll learn
100% online training 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 the intake, so you can combine it with your job.
Each module combines theory, guided reading of the original standard and applied exercises. The working software is Flower, NVIDIA FLARE, Opacus, OpenFHE and TensorFlow Federated, in free, academic or evaluation versions. Practical work is based on real procedures: federated training across hospitals, learning with differential privacy and evaluation of gradient reconstruction attacks. Assessment is continuous by module, plus the final Master's Thesis.
Standards covered: Regulation (EU) 2016/679, ISO/IEC 27701, Regulation (EU) 2022/868, ISO/IEC 29100, ISO/IEC 20889, Regulation (EU) 2024/1689, IEEE 3652.1.
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
- A university degree in engineering, science or architecture, or equivalent demonstrable professional experience.
- An internet connection and a computer capable of running the programme's 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 Flower, NVIDIA FLARE, Opacus, OpenFHE and TensorFlow Federated, 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 computer able to run the programme's software 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 Time Series, Signals and Edge AI Engineering
Online master's programme of 1,900 hours in time series, signals and edge AI: 12 modules covering discrete signals and systems, digital filtering, time-series analysis, deep learning…
Master’s in Advanced Vision/Geomatics Engineering (Beyond Mobility)
Online master's programme of 1,900 hours in advanced vision and geomatics engineering beyond mobility. 12 modules covering reference systems and geodesy, optical remote sensing, SAR and…
Master’s in Technical NLP and Engineering Assistants (RAG)
Online master's programme of 1,900 hours in technical NLP and engineering assistants (RAG) for documentation, logs and expert support. 12 modules covering natural language in engineering,…
More in Data, AI, Simulation & Analytics
All Data, AI, Simulation & Analytics programmes → · How to apply · Fees & financing · Scholarships