The current opportunity is available for current Manchester Met students, to apply you must be based in the UK for the duration of the role.
The main purpose of this role is to provide AI engineering and research support for CollabXAI: A Secure Software Platform to Enable Trusted Collaborative AI among SMEs. The successful candidate will contribute to the design, implementation and evaluation of the collaborative AI and trust-validation components of a TRL4 platform that enables SMEs to collaboratively train AI models without sharing their underlying data.
The initial demonstrator will focus on an intelligent intrusion detection system (IIDS) and will investigate privacy-preserving collaborative learning and mechanisms for identifying potentially unreliable or suspicious model updates.
The role will support the development of new approaches in collaborative/federated learning, trustworthy AI, AI security and privacy-preserving machine learning.
Dates: 5 October 2026 - 12 March 2027
Hours: 16 hours per week, with working pattern agreed with the project lead and subject to student working-hour restrictions.
Location: Manchester Metropolitan University, Manchester, with hybrid working where appropriate.
Key Tasks
Research Activities
- Contribute to the design, development and evaluation of the collaborative AI components of the CollabXAI TRL4 prototype.
- Develop and evaluate the initial intelligent intrusion detection system (IIDS) use case using appropriate benchmark/representative datasets, with opportunities to explore and develop 2–3 additional use cases, such as financial fraud detection, predictive maintenance, or supply-chain risk/anomaly detection, during the project.
- Prepare, preprocess and analyse data required for AI model development and experimental evaluation.
- Develop local machine-learning training workflows suitable for execution by CollabXAI agents within distributed SME environments.
Implement and evaluate collaborative/federated learning workflows, including appropriate baseline aggregation approaches such as Federated Averaging (FedAvg).
- Develop and evaluate the CollabXAI trust-validation mechanism for assessing model updates using evidence such as update deviation, cosine similarity, clipped norm bounds, validation-loss impact, timing and historical behaviour.
- Investigate approaches for incorporating trust scores into collaborative model aggregation and for identifying suspicious, anomalous or potentially poisoned model updates.
- Design and conduct controlled experiments involving benign and anomalous/malicious model updates to assess robustness and trust-validation behaviour.
- Evaluate collaborative AI performance using appropriate measures such as predictive performance, convergence, robustness, false-positive/false-negative behaviour and computational/communication overhead.
- Work with the Software Engineer to integrate AI training, model-update, trust-validation and aggregation components into the end-to-end CollabXAI platform.
- Undertake AI and machine-learning research using appropriate frameworks and tools such as ML.NET, Python, scikit-learn, PyTorch/TensorFlow and, depending on the final prototype requirements.
- Plan and prioritise own work and computational resources within the research project framework to achieve agreed research objectives and milestones.
Conduct literature and database searches relating to federated learning, collaborative AI, intrusion detection, privacy-preserving machine learning, model poisoning and trustworthy AI.
- Maintain reproducible experimental records and write up methods, results and technical findings for presentation to the research team and project stakeholders.
- Contribute to research reports, academic publications, demonstrations and dissemination materials arising from the project.
- Use initiative and judgement to develop appropriate AI techniques and resolve research and implementation problems affecting project objectives and deadlines.
Liaison and Networking
- Work closely with the Project Lead, Software Engineer, Pau&Co and participating SME partners to ensure that the AI components address the requirements identified through co-design and pilot activities.
- Present experimental results, AI demonstrations and research findings to the project team and relevant stakeholders.
Contribute, where appropriate, to academic/professional publications, conference outputs and project dissemination activities.
- Participate in relevant internal and external research partnerships and networks to share findings and good practice.
Service Provision
- Proactively and effectively engage with quality assurance and research integrity procedures to ensure that University standards are upheld.
- Collaborate with academic colleagues and project partners on research development, experimental evaluation and future directions for CollabXAI.
Team Working
- Actively participate as a member of the CollabXAI research team and work closely with the Software Engineer on end-to-end integration and testing.
- Maintain clear documentation of datasets, experiments, models, parameters, results and code to support reproducibility.
- Contribute to project meetings, technical reviews, demonstrations and pilot activities as required.