CS Top-Conference PhD Recruitment Roundup — Week of May 18, 2026

Five fully funded PhD positions confirmed this week, with two advisors who have published at ICML 2025 and one with an NeurIPS oral. Four of the five deadlines fall within the next four weeks. Geographic spread: Switzerland (2), Belgium (1), France (1), Sweden (1). Research directions span LLM reasoning, NLP for legal AI, causal ML, multimodal speech modeling, and computer vision for social good.

Deadlines at a glance

Deadline	Position	Institution	Direction
June 12, 2026	Multimodal Speech Language Models	INRIA Grenoble, France	Multimodal learning / NLP
June 13, 2026	Earth Observation + Deep Learning	Chalmers University, Sweden	Computer vision / AI for social good
June 18, 2026	Causal ML for Industrial Root Cause Analysis	KU Leuven, Belgium	Causal ML / safe AI
Rolling	Reasoning in Machine Learning (LLMs)	University of Basel, Switzerland	LLM reasoning / RL
Rolling	Applied ML & NLP (Responsible AI)	ETH Zurich, Switzerland	NLP / responsible AI

LLM reasoning and foundation models

Reasoning in Machine Learning — University of Basel, Switzerland

Supervisors Ilija Bogunovic and Aurelien Lucchi are looking for one PhD student to work on the foundations and applications of reasoning in large language models, including diffusion-based architectures 1.

The scope covers mathematical reasoning, RL-based policy optimization in complex environments, fine-tuning protocols for adapting pre-trained models to specialized domains, and closing generalization gaps. The project blends experimentation in few-shot learning and meta-learning with formal analysis. Publication targets are ICML, NeurIPS, ICLR, JMLR, and AISTATS.

Advisor background: Bogunovic had four papers at ICML 2025, including a spotlight, and is known for Bayesian optimization + LLM alignment work 2. Lucchi is an associate professor at Basel with 23,000+ Google Scholar citations and published at ICLR 2025 on adaptive optimizers under differential privacy 3.

Position details: 4-year PhD, fully funded, start ~July 2026. Basel has active collaborations with EPFL and ETH. Requires an MSc in mathematics, theoretical physics, or CS and strong coding skills. Applications are reviewed on a rolling basis; the post is open until filled.

NLP and responsible AI

Applied ML & NLP — ETH Zurich, Switzerland

ETH Zurich's Center for Law & Economics is hiring one PhD student for the SNSF-funded project Responsible AI for the Swiss Judiciary 4. The position sits at the intersection of ML, NLP, law, and empirical social science, developing and evaluating AI methods in judicial contexts.

Advisor background: Jakub Macina is an ETH AI Center Fellow whose doctoral work produced a NeurIPS 2022 oral presentation on neural dialog tutoring (co-authored with Iryna Gurevych at EACL 2023) 5. His current research focuses on LLM reasoning for high-stakes domains.

Position details: Fully funded, open until filled. Offices are in downtown Zurich. Requires strong ML/NLP skills and software engineering background. The role involves cross-disciplinary collaboration with legal scholars.

Lab culture note: The ETH AI Center fellowship structure gives student researchers access to the broader ETH/ETH AI Center network. The interdisciplinary framing of this project means applicants should be comfortable working with non-CS collaborators.

Causal ML and safe AI

Causal ML for industrial root cause analysis — KU Leuven, Belgium

Jesse Davis and Hendrik Blockeel at KU Leuven are recruiting one PhD student to investigate how causal ML can be leveraged for root cause analysis in multimodal industrial data 6. The project extends the CD-RCA framework, which uses causal discovery to estimate relationships between prediction errors and root causes.

Advisor background: Jesse Davis is a full professor in KU Leuven's CS department with 18,000+ citations. He had a paper in ACL 2025 Findings (with Marie-Francine Moens' LIIR lab) 7 and has sustained publication output at ICML, AAAI, and related venues. His research spans machine learning, data mining, and AI safety through tree-ensemble verification.

Position details: Fully funded, deadline June 18, 2026. Location: Leuven, Belgium. Requires MSc in CS or a closely related field; ML background is expected.

Multimodal learning and speech

Multimodal speech language models for language acquisition — INRIA Grenoble, France

Thomas Hueber (GIPSA-lab, CNRS / Université Grenoble Alpes) is leading a PhD on how multimodal and social interactions contribute to human language acquisition 8. The project builds on textless Speech Language Models (SpeechLMs) and developmental AI paradigms inspired by how infants learn language, with co-supervision by Stéphane Lathuilière and Laurent Girin.

Advisor background: Hueber has publications in the ACL Anthology including EMNLP 2025, as well as papers at Interspeech 2022 and earlier 9. Google Scholar shows 3,600+ citations; research spans automatic speech processing, visual speech recognition, and ML for assistive technology.

Position details: 3-year PhD, fully funded, salary ~€2,300 gross/month (standard French doctoral contract). Deadline June 12, 2026. Location: Grenoble, France. The position is funded by the DevAI&Speech Chair of the MIAI Cluster. Requires MSc in engineering or CS with ML and signal processing background.

Earth observation and AI for poverty estimation — Chalmers University of Technology, Sweden

Chalmers' Department of Computer Science and Engineering is recruiting one doctoral student to develop deep-learning methods that estimate multidimensional poverty from Sentinel-2 satellite imagery across Africa 10. The project has three objectives: build reliable poverty estimation models from satellite time series (quarterly, 1984–2025), characterize temporal dynamics of living conditions, and produce deployable tools for SDG monitoring. Research from this group has already appeared in peer-reviewed journals 11.

Note on supervisor verification: The lead supervisor for this position is not named in the public job posting; the research group has relevant publications in satellite-AI for poverty estimation, but the PI's top-venue ML publication record was not independently confirmed this week. Applicants interested in this direction should review the group's publication list before applying.

Position details: Fully funded doctoral position (Swedish model, ~4–5 years). Deadline June 13, 2026. Location: Gothenburg, Sweden. Requires a background in ML/deep learning; experience with satellite or earth observation data is an advantage.

Notes for applicants

Two deadlines in the next 10 days: INRIA Grenoble (June 12) and Chalmers (June 13) are the most time-sensitive openings this week. The KU Leuven causal ML position closes June 18.
Switzerland concentration: Both Basel and ETH Zurich positions are rolling and open until filled. Given the ICML and NeurIPS pedigree of both advisory teams, early contact is advisable rather than waiting for a formal deadline.
Research direction gaps: No positions this week in reinforcement learning, ML theory, or robotics. US-based lab announcements remain absent from this channel's current sourcing approach; the centralized application model at most US PhD programs makes individual advisor postings harder to surface.
Lab culture information was available only for the ETH Zurich position (ETH AI Center network access, interdisciplinary framing) and the Basel position (collaborations with EPFL and ETH). For the remaining positions, applicants should consult lab websites and reach out directly to current PhD students before applying.