Overview · Research collaboration
Fund a study, collaborate, or exchange methods
There are five ways to work with me on research. They differ in who pays, who owns what is produced and whether the result is published, so it is worth choosing the route before writing the brief.
The situation
We have funding for a defined technical uncertainty. How can we structure a sponsored study with useful deliverables and credible methods?
What you leave with
A route that fits your funding and publication constraints, then either a scoped sponsored study or a selective, explicitly non-commercial collaboration.
For: Principal investigator; R&D lead; consortium coordinator
Choosing a route
| Route | Who funds | Who owns outputs | Publication default | Typical output |
|---|---|---|---|---|
| Sponsored study | The sponsor | Agreed at scoping; each side keeps its background IP | Confidential unless publication is agreed in advance | Protocol, raw results, report and decision memo |
| Open research experiment | The sponsor, by milestone | Openly licensed code and write-up; data where licensing allows | Published, whatever the result | Preprint, public artefacts, negative results included |
| Industry–academic collaboration | Each party funds its own contribution, or a grant held by the institution | Per the collaboration agreement and institutional policy | Published | Joint paper, authorship by contribution |
| Consortium technical contribution | Per the consortium’s own arrangements | Per the consortium agreement | Per the consortium agreement | Evaluation design or technical work-package input |
| Seminar or methods exchange | Non-commercial | Slides and materials stay with their authors | Recording and slides only if the host and I agree | Abstract, reading list, discussion |
Two distinctions matter more than the table:
- Paid or not. Work here is either sponsored — a fixed scope with a fee that does not change with the outcome — or explicitly non-commercial. There is no middle category where an informal collaboration quietly becomes a commercial deliverable.
- Who decides publication, and when. For every route, the publication decision is made before results exist. A study whose publication depends on whether the result is favourable is not one I can run credibly.
Which route fits
- You have a budget and need an answer for an internal decision → a sponsored study. Start with the study brief.
- You want the answer to be public and reusable by others → sponsor an open experiment.
- You are a principal investigator with a shared research question → industry–academic collaboration.
- You coordinate a funded consortium and need a technical contributor for a defined work package → consortium contribution. I hold no institutional signing authority, so eligibility and contracting run through the consortium’s own process.
- You run a seminar series or reading group → academic seminars.
Whichever route you choose, publication, IP and independence covers background and foreground IP, review windows, authorship and conflict disclosure. Read it before committing funds.
Useful if
- You have a technical uncertainty in AI evaluation, generated-code correctness, federated or privacy-preserving learning, or decentralised systems, and want it investigated with stated methods.
- You can say in advance how ownership, authorship and publication should work, or are willing to settle them before any work starts.
- You accept that the answer may be that the approach does not work.
Not the right fit if
- You need a system built, fixed or operated. That is delivery work, not research.
- You want a conclusion endorsed rather than tested.
- You are looking for a keynote or conference talk; that is a separate, bookable event product.
What you need before starting
- The question. One or two sentences on the technical uncertainty and why it matters now.
- The funding position. Whether money exists for this, and from where: a company budget, a grant held by an institution, a consortium work package, or none.
- Publication constraints. Whether results must stay confidential, must be published, or can be decided later — with the decision made before any results exist.
Evidence behind this page
- How Reproducible Are Evaluation Conclusions? A Self-Audit of LLM-Inferred Prompt Structure
2026 · Preprint (not peer reviewed)
- The Correctness Illusion in LLM-Generated GPU Kernels
2026 · Preprint (not peer reviewed) · cited by 6 (Google Scholar, October 2026)
- Fed-Focal Loss for imbalanced data classification in Federated Learning
2020 · Peer-reviewed workshop paper · cited by 96 (Google Scholar, October 2026)
Narrower routes from here
Sponsored research
Write a Sponsored AI Research Brief
Structure a sponsored AI study around a claim, a decision, a baseline, realistic access and stop criteria.
Sponsored research
Publication, IP and Research Independence
A discussion checklist for IP, data handling, publication review, unfavourable results, authorship and conflict disclosure.
Sponsored research
Sponsor an Open Research Experiment
Fund a public, milestone-based experiment with open methods and artefacts, published whatever the result.
Collaboration
Industry-Academic AI Research Collaboration
Map a shared research question, split contributions and settle data and publication dependencies before the work starts.
Collaboration
Technical Contributions to Research Consortia
Scope a bounded technical work package for a research consortium, with dependencies, deliverables and an honest capacity statement.
Collaboration
Academic Seminars and Methods Exchanges
Research seminars and methods discussions for labs and seminar series, with an abstract, prerequisite reading and discussion questions.
Questions
What is the difference between a sponsored study and a research collaboration?
A sponsored study is paid work with a fixed scope and a fee that does not depend on the result. A collaboration is explicitly non-commercial: each party contributes time or resources, and authorship follows contribution. Interest in collaborating does not create a commercial engagement, and a collaboration is not unpaid consulting.
Can an industry team commission applied AI research and still publish it?
Yes, if publication is agreed before the study starts. Publication review, ownership of what is produced and the handling of unfavourable results are settled at scoping; the publication, IP and independence page lists what to agree.
Do you take on every collaboration request?
No. Collaborations are selective and depend on fit with my current research questions and on capacity. A sponsored study is assessed on whether the question is answerable with the available access.
Are there conflicts of interest I should know about?
I founded Neul Labs, which builds AI agent infrastructure in Rust, and my research affiliation is Skelf Research. If I have built, advised or hold an interest in a system relevant to the question, that is disclosed at intake and may mean declining or narrowing the work.
See also
Send a research brief
Describe the uncertainty, the funding position and any publication constraint. A short, non-confidential note is enough; I will suggest the route that fits, or say if none does.
Last reviewed 2026-10-07.