I am Pontus Stenetorp (legally, Pontus Lars Erik Saito Stenetorp or 西東 ステネートルプ ポントス ラース エリック) and I find computers, science, and technology in general to be fascinating.
At University College London (UCL), I am a Professor of Natural Language Processing (NLP) in the Department of Computer Science and, among other duties, lead the NLP Group, am a Deputy Director for the Centre for Artificial Intelligence, and Programme Director for the Facebook Artificial Intelligence/UCL PhD Programme. In addition, at the National Institute of Informatics (NII; 国立情報学研究所) I am a Specially Appointed Researcher (特任研究員) in the Research and Development Center for Large Language Models (LLMC; 大規模言語モデル研究開発センター).
In terms of research, my expertise and interests lie in the intersection of human language and computers. The great majority of my research is empirical in nature and on a high level explores questions such as "How can we improve a computer's ability to process language?" and "What are the current limits of a computer's ability to process language and how do we best measure this empirically?". For an overview of the research I conduct with students, researchers, and collaborators please have a look at my list of publications.
Frequently asked questions
"Can I apply for PhD with you as my primary supervisor?"
As primary supervisor, I am not recruiting PhD students for the 2024/25 and 2025/26 academic years. However, do consider applying with my colleagues Aldo Lipani and Emine Yilmaz, as I am still open to acting as a subsidiary supervisor if the research direction aligns with my own. Additionally, you can apply for the Facebook Artificial Intelligence/UCL PhD Programme where I regularly act as the primary academic supervisor.
"XXX: FAIR PhD!"
XXX:
"Can I do an internship with you?"
I offer neither in-person or remote internships for a number of reasons. Firstly, the necessary paperwork is both complicated and takes a very long time to process. Secondly, internships are generally too short to accomplish anything concrete enough to be published. Thirdly, my workload does not allow for additional supervision.
"Can I do a research visit with you?"
Maybe, if there is an ongoing research collaboration with you or your institution, you have previous experience conducting and publishing academic research, and strong alignment with my research. For the great majority of requests however the answer is no, for the same reasons as for internships.
"Do you offer research assistant positions?"
I do not offer paid or unpaid research assistant positions as my workload does not allow me to take on additional managerial duties.
"Can you supervise my BSc/MEng/MSc thesis?"
Each year I supervise over a handful of MSc theses. The great majority of which are for the Machine Learning and Data Science and Machine Learning MSc programmes. Throughout the terms, the supervised students are embedded within the NLP Group and we pursue projects which are closely aligned with current research. While the department formalises supervisor assignment in the second term, do reach out early if you are interested as I seldom have the capacity to meet the demand for thesis supervision.
Due to my workload, I only supervise BSc, MEng, or external theses under exceptional circumstances: such as the student already having a strong understanding and experience with my area of research.
I rarely supervise "industry projects", where the student mainly will work with an industrial partner, unless there is a very strong research component and the goal of the project is academic research rather than product oriented.
To be considered for a project, please send me an up-to-date CV and brief description (not more than three paragraphs) of what kind of project you are interested in (preferably relating it to my research).
"Can I take COMP0087: Statistical Natural Language Processing?"
XXX: If elective...
"Can you write me a letter of recommendation?"
For students, yes, if we have worked closely together for some time so that I have enough experience working with you to form the basis of the letter. Note that taking a module that I am giving is usually insufficient as the most interesting I can then write in the letter would be "They did well in my module", which is highly unlikely to impress whoever reads it.
When it comes to fellow academics, I am happy to write letters of recommendation as long as I am familiar with your research. A good indicator would be if my work cites yours, but do reach out as I am happy to help as long as I have enough to write about you and your work.
Contact
I prefer e-mail over all other modes of communication and process approximately one hundred e-mails per day. While I have more than two handfuls of e-mail addresses, they are all forwarded to a unified inbox and thus you will only ever need to send a single e-mail to one of my e-mail addresses for it to reach me. The great majority of academic queries that I receive can be answered by the frequently asked questions section and I reserve the right not to respond to e-mails or respond with a simple link to the relevant part of the section if your query can be answered by it.
Written, asynchronous communication can be challenging, and I ask that you follow a few "best practices" for e-mail: to make it easy for the receiver to understand the content swiftly, write a subject that captures your query in its entirety and summarise your message at the top of your e-mail; to make following long-running conversations easy, do not top post (either bottom posting or interleaved replies are perfectly fine); use plaintext e-mail; to allow for offline access, attach documents rather than link them; and preferably avoid sending reminders as I process older e-mail chains first and thus reminders will effectively delay your response (if you absolutely must send reminders, please wait at least a work week before sending one).
I use extensive, automated filtering to triage incoming e-mail based on, among other things, which e-mail address it was sent to. For matters related to UCL, please use p.stenetorp@cs.ucl.ac.uk (An e-mail address ending in @ucl.ac.uk is erronously listed on some UCL websites as belonging to me. Please do not use it, as e-mails sent to it will not reach me.) and for matters related to NII, please use pontus@nii.ac.jp. If your query is academic in nature but not directly related to a specific institution, either the UCL or NII address will work just fine. For all other matters, please use pontus@stenetorp.se.