Location: Dublin Innovation Institute, Block 6 Founders District, Belfield Office Park, Beech Hill Road, Dublin 4, Ireland

Working arrangement: Full Time on site at the Dublin Innovation Institute

Click to read more about the Dublin Innovation Institute (DII): https://dublininstitute.northeastern.edu/about/

Term: 10 months, full time

Reports to: Dr. Shashi Murthy, Head of Institute. Prof. Giovanni Petri

Start: 1 September 2026 or as soon as possible thereafter

The role
Northeastern University’s Dublin Innovation Institute (DII) wishes to appoint a Postdoctoral Research Associate to work on TOPO-LLM: Topologically-Informed Large Language Models for Innovation Intelligence. The project is funded by FFplus, a EuroHPC Joint Undertaking initiative under the European Union’s Digital Europe Programme, and brings together DII and Futurity Systems, a Barcelona-based technology company developing AI tools for innovation and competitive intelligence.

TOPO-LLM asks whether large language models can reason more effectively about scientific and technological innovation when they are given explicit information about the higher-order structure of the underlying data. The project will combine topological data analysis, higher-order network methods and large-scale language-model training on a knowledge graph of scientific publications, patents and related innovation data.

The appointed researcher will lead DII’s scientific contribution. The main task is to design and validate an Innovation Topology Embedding: a compact representation built from persistent-homology summaries, higher-order Laplacian features and multiscale network structure. The researcher will develop scalable methods for computing these features on a very large knowledge graph, integrate them into a transformer-based language model, and test their value through carefully controlled benchmarks and ablation studies.

This is a deliberately interdisciplinary role. We are looking for someone who combines a strong foundation in topological or geometric methods with the ability to implement and evaluate machine-learning systems at scale. Candidates may come from topological data analysis, geometric deep learning, network science, statistical physics, machine learning, or a closely related field.

We recognise that few applicants will be equally strong in every area; where a candidate’s experience is weighted towards either topology or machine-learning engineering, we particularly value evidence that they can collaborate across the boundary and learn quickly.

The researcher will be based at DII on the University College Dublin campus and will join Prof. Giovanni Petri’s research group, NP Lab, within Northeastern’s Network Science Institute London. They will work closely with the Futurity Systems team in Barcelona and with collaborators across Northeastern’s global network. The project includes access to the EuroHPC MareNostrum 5 supercomputer at the Barcelona Supercomputing Centre.

We provide an opportunity to carry out academically substantive research in close contact with an industry partner. The successful candidate will be expected to contribute to research publications, project reports and reusable research software arising from the work, subject to the project’s collaboration and intellectual-property agreements.

We know diversity fosters creativity and innovation. We are committed to equality of opportunity, to being fair and inclusive, and to being a place where all belong. We therefore particularly encourage applications from candidates who are likely to be underrepresented in the applicant pool.

Dublin Innovation Institute and the research group

Northeastern University’s Dublin Innovation Institute serves as a bridge between the University’s global network and the European research ecosystem. Located on the campus of University College Dublin, DII offers opportunities for collaboration across Europe and North America. Its current projects span political science, computer science and network science, and this appointment will form part of the Institute’s initial cohort of research staff. The position will be part of NPLab, led by Prof.

Giovanni
Petri, Professor of Network Science at Northeastern University London and core faculty at the Network Science Institute. The group works at the intersection of higher-order network science, topological data analysis and information theory, with applications in machine learning, neuroscience, collective behaviour and animal communication. NPLab brings together doctoral and postdoctoral researchers in London and across Northeastern’s global network.

This post is funded by FFplus. The research directly relevant to the position sits within the group’s wider work on higher-order Laplacian renormalisation, higher-order connectomics, and topological representations of complex systems. Prof.

Petri leads a broader research programme in this area, including as Principal Investigator on the ERC Consolidator Grant RUNES and Project CETI, and as Co-Investigator on the BeyondTheEdge Marie Skłodowska-Curie Doctoral Network, giving the group a rich international research environment beyond this specific project.

Network Science Institute
Northeastern University’s Network Science Institute (NetSI) brings together researchers from the physical, information, biological and social sciences to develop the theory and applications of complex networks. Founded in Boston in 2014, NetSI now connects research communities across Northeastern’s global network, including its London hub. The successful candidate will join this international and highly interdisciplinary community.

About Northeastern University
Founded in 1898, Northeastern is a global research university and a recognised leader in experience-driven lifelong learning. Its locations across the United States, Canada, the United Kingdom, and Ireland create opportunities for collaborative, solutions-focused research across disciplines and regions.

Northeastern University is an equal opportunity employer, seeking to recruit and support a broadly diverse community of faculty and staff. Northeastern values and celebrates diversity in all its forms and strives to foster an inclusive culture built on respect.

Job Duties The appointed researcher will carry out independent research, analyse and publish results, prepare project reports and technical documentation, and contribute to the wider research environment at DII and NPLab.

Job duties will include the following, with the distribution of time varying across the project:

60-70% Research and implementation: design the Innovation Topology Embedding; develop and scale the topology-computation pipeline; integrate topological features into the language model; and run controlled evaluations and ablation studies. 15-20% Research communication: prepare publications, project deliverables, presentations and documentation for research software. 10-15% Project collaboration: coordinate technical work with Futurity Systems, manage use of the EuroHPC allocation, and report progress against project milestones. 5-10% Research-group activity: participate in NP Lab and Network Science Institute meetings and contribute to the group’s collaborative research culture. Person Specification Criteria A PhD, awarded or near completion by the start date, in computer science, mathematics, physics, statistics, network science, or a related quantitative discipline. Research experience in at least one of the following: topological data analysis; computational topology; geometric or topological deep learning; higher-order network science; graph machine learning; or statistical physics of complex systems. Strong programming skills in Python and experience with modern scientific or machine-learning software. Experience designing, running and interpreting computational experiments. Demonstrated ability to plan, execute and publish independent research. Ability to communicate technical ideas clearly in written and spoken English. A collaborative working style, strong personal initiative and the ability to deliver against project milestones. We recognise that excellent candidates may not match every item listed above. If your background is adjacent and you are excited by the project, we encourage you to apply.
Additional desired skills and experience:

Practical experience with persistent homology or related topological descriptors, using tools such as GUDHI, Ripser, giotto-tda, or equivalent. Experience with transformers or large language models, including fine-tuning, continued pretraining, representation learning, or model evaluation. Experience with PyTorch and the Hugging Face ecosystem. Experience with high-performance computing, SLURM, distributed multi-GPU training, or large-scale checkpointing. Familiarity with hypergraphs, simplicial complexes, higher-order Laplacians, network renormalization, or multiscale graph methods. Experience with large knowledge graphs, sparse linear algebra, graph databases, or bibliometric and patent data. A track record in open-source scientific software or research collaborations with external partners. Additional Information
Enquiries
Informal enquiries may be made to Prof. Giovanni Petri at and are warmly encouraged. To be considered for the position, applications must be made in accordance with the process specified in the final advertisement.

Application Process The panel will be shortlisting for this position on a rolling basis so please apply as soon as possible. We reserve the right to close this post before the closing date if we receive a large number of applications.

Applications should be made to Caroline Ward, HR Director () by 12pm on Wednesday 09 September 2026. Please reference TOPO-LLM-DII-2026 in your application.

Please ensure that your application includes the following:

A CV, including a publication list and details of relevant software or open-source contributions. A covering letter explaining how you meet the selection criteria and why you are interested in the role. A research statement of no more than two pages and at least 1 figure, describing your relevant work and how you would approach the first six months of the project. Names and contact information for up to three referees. References will only be sought for shortlisted candidates. Shortlisted candidates will be interviewed by video call. The interview will include a technical discussion of the candidate’s relevant work and of the practical challenges involved in scaling topological methods and language-model training.
Eligibility to work. Due to DII’s current staffing requirements, at this time we are only able to appoint an EEA, UK or Swiss national to this role. Applicants must also be able to demonstrate their entitlement to work in Ireland. Please describe your status in your covering letter.

Postdoctoral Research Associate in Topological Machine Learning and Generative AI

About this role

Company:

Qualifications:

Language requirements:

Specific requirements:

Educational level:

Level of experience (years):

Similar roles