ML/NLP PhD & Postdoc positions — May 2026 digest
The inaugural weekly ML/NLP positions digest. One fully confirmed hard-deadline opportunity at CopeNLU (University of Copenhagen) closes May 31, 2026: a 3-year PhD fellowship in Mechanistic Interpretability for LLM Security and a 3-year Postdoc in LLM Factuality Detection under Isabelle Augenstein and Pepa Atanasova. Two rolling-admission labs (Schütze Lab/LMU Munich and McGill NLP/Mila) also included.
This is the inaugural edition of a weekly digest tracking open PhD and Postdoc positions from PIs with strong publication records at top ML/NLP venues (ACL, EMNLP, NAACL, NeurIPS, ICML, ICLR, and related conferences). Each position is verified against official lab or university pages before inclusion.
One position closes in four days (May 31). If you are reading this in the week of May 27, act on the CopeNLU listing immediately.
Confirmed open positions
CopeNLU — University of Copenhagen
Lab overview: CopeNLU is the Natural Language Processing research group at the University of Copenhagen's Department of Computer Science, affiliated with the Pioneer Centre for AI (a Danish national AI research hub). The group consistently ranks among Europe's most productive NLP labs, with 6 papers accepted to ACL 2026, 8 to EMNLP 2025, 3 to ACL 2025, 4 to NAACL 2025, and 5 to EMNLP 2024, plus an Outstanding Paper Award at EACL 2024. 1
Principal Investigators:
- Isabelle Augenstein — Full Professor; research focuses on tasks requiring deep language understanding rather than surface-level pattern matching, including learning from limited labeled data, explainable AI, fact verification, gender bias detection, and question answering.
- Pepa Atanasova — Assistant Professor; research focuses on developing, diagnosing, and applying interpretability and explainability techniques for NLP models.
Funding project: "A Mechanistic Framework for Mitigating the Susceptibility of LLMs to Learning False Information," funded by the Danish Independent Research Fund (DFF). Both positions include access to NVIDIA academic collaboration resources. 2
Position 1: PhD fellowship in Mechanistic Interpretability for LLM Security
| Field | Details |
|---|---|
| Position type | PhD fellowship (full-time, 3 years) |
| Start date | 1 September 2026 |
| Location | Copenhagen, Denmark |
| Research focus | Mechanistic interpretability methods to prevent LLMs from internalizing false information across lifecycle stages |
| Required qualification | Master's degree in Computer Science or a closely related field |
| Preferred background | ML/NLP research experience |
| Deadline | 31 May 2026, 23:59 GMT+1 |
| Salary/funding | Fully funded (DFF grant) |
| Apply | Official application portal |
正在加载内容卡片…
Required application materials:
- Cover letter
- Research statement (describing your specific research direction and preliminary methodology)
- CV
- Degree certificate(s) and transcripts
- Publications list (if applicable)
- Contact details for three referees
Position 2: Postdoc in LLM Factuality Detection
| Field | Details |
|---|---|
| Position type | Postdoctoral researcher (full-time, 3-year contract) |
| Start date | 1 September 2026 |
| Location | Copenhagen, Denmark |
| Research focus | Characterizing the spectrum of false information across LLM lifecycle stages; developing mitigation methods |
| Required qualification | PhD in Computer Science or a closely related field |
| Deadline | 31 May 2026, 23:59 GMT+1 |
| Salary/funding | Fully funded (DFF grant) |
| Apply | Official application portal |
正在加载内容卡片…
Required application materials:
- Cover letter
- Research plan
- CV
- PhD degree certificate
- Complete publications list
- 3 representative papers
- Contact details for three referees
Rolling-admission labs worth contacting now
These labs accept PhD and Postdoc applications on a rolling basis — there is no fixed public deadline, but earlier contact typically increases your chances of being reviewed in the next admission cycle.
CIS Lab — LMU Munich (Schütze Lab)
PI: Hinrich Schütze — Professor; one of the most cited researchers in NLP (ACL Fellow). The lab had 6 papers at EACL 2026 and 6 at NAACL 2025. Research covers multilingual NLP, low-resource learning, representation learning, and computational semantics. 5
How to apply: Send your CV directly to [email protected]. No fixed deadline; rolling review. No specific advertised positions are currently listed — the lab accepts unsolicited PhD and Postdoc applications when a strong fit is identified.
| Field | Details |
|---|---|
| Position types accepted | PhD, Postdoc |
| Institution | Ludwig Maximilian University of Munich (LMU Munich), Germany |
| Deadline | Rolling (no fixed date) |
| Contact | [email protected] |
McGill NLP — McGill University / Mila
Lab overview: McGill NLP brings together multiple PIs affiliated with both McGill University and Mila (Quebec AI Institute), one of the world's leading academic ML research centers. The group covers a broad range of NLP and ML research directions. 6
How to apply: PhD and MSc admissions go through the Mila + McGill joint admission system — applications are reviewed on a rolling basis. Postdoc applicants should contact individual PIs directly. The lab page does not list specific open positions; check the Mila application portal and individual PI lab pages for current intake.
| Field | Details |
|---|---|
| Position types accepted | PhD, MSc, Postdoc |
| Institution | McGill University / Mila, Montreal, Canada |
| Deadline | Rolling (via Mila + McGill system) |
| Contact | Individual PIs via lab page |
Lab with no current openings
Rycolab — ETH Zurich (PI: Ryan Cotterell, Professor): As of the date this digest was researched, Rycolab is not accepting PhD students. The lab's official page states this explicitly. 7 MSc thesis supervision for current ETH students and potential co-supervision arrangements may still be possible — contact the lab directly if that applies to your situation.
Application guidance
Fit the research statement to the project, not the lab broadly
Both CopeNLU positions are funded under a specific DFF grant on mechanistic interpretability and LLM factuality. A generic "I'm interested in NLP" statement will not be competitive. Your research statement should engage with one or more of these angles:
- Mechanistic interpretability methods (circuits, probing, causal tracing, activation patching)
- How LLMs acquire, store, and propagate false or misleading information
- Factuality detection, fact verification, or hallucination mitigation
- Connections to adversarial robustness or model security
PhD vs. Postdoc at CopeNLU: what the research plans need to show
The PhD fellowship asks for a "preliminary methodology" — you do not need a complete research plan, but you should show you have thought through one concrete approach and why it fits the project scope. The Postdoc position asks for a full "research plan" and three representative papers, which signals the hiring committee expects a demonstrated track record of independent research.
Deadline logistics for CopeNLU
The deadline is 31 May 2026 at 23:59 GMT+1 (midnight Copenhagen time). That is equivalent to:
- 23:59 CEST (Central European Summer Time)
- 17:59 EDT / 14:59 PDT on May 31
Both positions use the University of Copenhagen's HR-Manager portal. The portal typically requires creating an account before uploading documents — allow time for account setup if you have not used the system before.
Cover image: AI-generated illustration
参考来源
- 1CopeNLU homepage
- 2Funded PhD and Postdoc positions for start in Autumn 2026
- 3PhD fellowship in Mechanistic Interpretability for LLM Security — University of Copenhagen
- 4Postdoc in LLM Factuality Detection — University of Copenhagen
- 5CIS Lab / Schütze Lab — Vacancies
- 6McGill NLP — Join us
- 7Rycolab — ETH Zurich
围绕这条内容继续补充观点或上下文。