CS PhD Roundup — Week of June 22, 2026

Coverage window: June 15–22, 2026. Three new positions qualify this week — a PhD at INRIA Paris under Cordelia Schmid (robotics/computer vision), a postdoc at SDU Denmark under Melih Kandemir (neuromorphic RL), and two postdocs at EPFL under Charlotte Bunne (biological foundation models and frontier generative AI). All four positions from last week's imminent-deadline cluster — Vienna, UiT, KU Leuven, and INRIA Sophia Antipolis — have now closed. The three rolling positions at ISTA (Lampert and Locatello groups) and University of Bonn (Rußwurm MEO Lab) remain open with no change in status.

At a glance — all active positions

Deadline	Advisor(s)	Institution	Sub-field	Type
July 17, 2026	Cordelia Schmid	INRIA Paris, WILLOW team	Robotics / Computer Vision	PhD
September 1, 2026	Melih Kandemir	SDU ADIN Lab, Denmark	Neuromorphic RL	Postdoc
Rolling	Charlotte Bunne	EPFL AIMM Lab, Switzerland	Biological foundation models	Postdoc (×2)
Rolling	Charlotte Bunne	EPFL AIMM Lab, Switzerland	Frontier generative AI	Postdoc
Rolling	Christoph H. Lampert	ISTA, Austria	Trustworthy ML, neuro-symbolic AI	Postdoc (×2)
Rolling	Francesco Locatello	ISTA, Austria	Causal learning and AI	PhD & Postdoc
Rolling	Marc Rußwurm	University of Bonn, Germany	Geospatial neural fields	PhD (×2)

New this week — robotics and computer vision (PhD)

PhD in structured ego-world models — Cordelia Schmid, INRIA Paris (WILLOW team)

Deadline: July 17, 2026 · Start: September 1, 2026

INRIA Paris has posted PhD position 2026-10180, supervised by Cordelia Schmid (Research Director, INRIA; Research Scientist, Google; head of the WILLOW team). 1 The research focus is on building 3D/4D geometric world models that can transfer dexterous manipulation skills across robot embodiments — specifically, models that decouple how the environment behaves from how a robot's own body moves.

The three-year program runs in sequence: Year 1 targets latent representations for geometric consistency, temporal scene memory, and affordance modeling. Year 2 develops factorized world models with separate dynamics streams for the world and for the robot's kinematics, enabling cross-embodiment transfer from human video to robot policies. Year 3 covers cross-embodiment adaptation and the design of new benchmarks for transfer robustness. 1

The motivation framed in the listing: current state-of-the-art robot learning policies are "severely brittle to cross-embodiment shifts, viewpoint perturbations, out-of-distribution visual environments, and the compounding errors inherent to long-horizon inference," with the core bottleneck being "a monolithic, direct pixel-to-action mapping." 1 The proposed direction is structured 3D/4D representations and factorized world models grounded in geometry rather than raw pixels.

Schmid's top-venue output (2022–2025): CVPR 2024 (8+ papers, including Pixel Aligned Language Models, MoReVQA, Streaming Dense Video Captioning, SUGAR, End-to-End Spatio-Temporal Action Localisation); CVPR 2025 (7+ papers, including FirePlace, Visual Lexicon, InteractVLM, Language-Guided Image Tokenization); ICLR 2024 and 2025; ICML 2024; NeurIPS 2024; ACL 2025 Findings. 2 She received the Körber European Science Prize in 2023 and holds a dual appointment with Google Research.

Robotic arm grasping a geometric object on a lab bench, holographic wireframe overlays in cyan-green, dual monitors showing 3D scene reconstruction, tungsten lamp lighting — AI-generated illustration of a dexterous robot learning lab — visual reference for the ego-world model research context.

Security note: The position is located in a Zone à Régime Restrictif (ZRR). An adverse ministerial decision on the required French government clearance cancels the appointment. Applicants who are not French nationals should contact INRIA before investing significant time in the application. 1

Field	Details
Position type	PhD (3-year contract)
Institution	Centre Inria de Paris, WILLOW team
Advisor	Cordelia Schmid (also Research Scientist, Google)
Stipend	€2,300/month gross
Start date	September 1, 2026
Location	Paris, France
Security	ZRR clearance required
Deadline	July 17, 2026
Requirements	Master's in CS, computer vision, robotics, or ML; programming proficiency; English required
Apply	jobs.inria.fr — offer 2026-10180 · contact: cordelia.schmid@inria.fr

New this week — neuromorphic reinforcement learning (postdoc)

Postdoc in neuromorphic RL — Melih Kandemir, SDU ADIN Lab (Denmark)

Deadline: September 1, 2026 · Earliest start: October 1, 2026

The ADIN Lab (Adaptive Intelligence Lab) at the University of Southern Denmark (SDU, Odense) is recruiting a two-year postdoc in neuromorphic reinforcement learning. 3 Supervisor is Prof. Melih Kandemir, who has published at NeurIPS 2022, 2023, and 2024 and at ICLR 2022. 4

The research agenda sits at the intersection of deep RL, probabilistic modeling, and spiking neural networks (SNN). The stated goal is adaptive control in non-stationary environments with applications to robotic systems and embodied AI — specifically, providing "theoretical guarantees (e.g., convergence, stability, or sample complexity) for control tasks in non-stationary environments." 3 The lab's framing is that the boundary between training and deployment should not be fixed — the agent adapts throughout its operational life.

The position carries standard Danish academic collective-agreement compensation and full Danish social benefits. SDU's main campus is in Odense, Denmark.

Spiking neural network rendered as a brain-shaped luminous graph with amber and violet pulsing nodes on deep indigo background — AI-generated illustration of a spiking neural network — conceptual reference for the neuromorphic RL direction.

Application requirements (hard): Completed PhD in CS, mathematics, statistics, or theoretical physics. At least two first-author papers accepted at NeurIPS, ICML, ICLR, or AISTATS. Proficiency in Python/PyTorch/JAX. 3 The publication bar is explicit and non-negotiable by the listing's own wording.

Field	Details
Position type	Postdoc (2-year contract)
Institution	SDU ADIN Lab, Odense, Denmark
Advisor	Prof. Melih Kandemir
Salary	Danish academic collective agreement rate (competitive postdoc salary + Danish social benefits)
Start date	October 1, 2026 (earliest)
Location	Odense, Denmark
Deadline	September 1, 2026
Requirements	PhD in CS / math / stats / theoretical physics; ≥2 first-author papers at NeurIPS / ICML / ICLR / AISTATS; Python / PyTorch / JAX
Apply	SDU Oracle Cloud recruitment system, Job ID 4085 — CV + full publication list + PhD certificate

New this week — biological foundation models and frontier AI (postdoc ×3)

Three postdoc positions — Charlotte Bunne, EPFL AIMM Lab (Switzerland)

Deadline: Rolling — priority review for fall 2026 start

The AIMM Lab (Artificial Intelligence in Molecular Medicine) at EPFL, led by Prof. Charlotte Bunne, has two distinct postdoc openings posted this month. Bunne holds appointments in EPFL's School of Computer Science, School of Life Sciences, and the EPFL AI Center. Her lab has published at NeurIPS (2022). 5

Position A — biological foundation models (2 postdocs) 5

The listing describes the scope as "the full stack of biological foundation model research: from core architecture design and pretraining at scale to integration into agentic interfaces." Research directions include pretraining objectives, tokenization for biological data, scalable architectures, multi-modal learning across biological data types (genomics, imaging, proteomics), and translational applications in drug discovery and biomedicine. Active industry partnerships with pharmaceutical companies are part of the research context. Stipend: competitive EPFL postdoc salary plus full social benefits. Compute: allocations on the Swiss National Supercomputing Centre (CSCS) and the Swiss AI Initiative.

Requirements: ML/CS/computational biology PhD; demonstrated large-scale AI development experience (pretraining, fine-tuning, scaling); familiarity with biological data modalities; strong publication record at top ML venues or life sciences journals.

Apply: Send CV, one-page research interest statement, and two references to bunnelab@gmail.com with subject line [AIMM Lab] Application PostDoc. Also available to apply through the EPFL careers portal.

Position B — frontier AI research / generative AI (1 postdoc) 6

A separate, methodology-forward role. The listing states: "The position is highly methodological and focuses on developing new generative modeling approaches, scalable training algorithms, and foundation model technologies." 6 Methods targeted include diffusion models, flow matching, and generative Transformers. No biology or biomedical background required.

Requirements: PhD in ML, CS, applied mathematics, statistics, physics, or engineering; demonstrated generative AI expertise; large-scale and distributed training experience; strong top-venue publication record.

Apply: Same path as Position A (bunnelab@gmail.com, [AIMM Lab] Application PostDoc).

Protein structure ribbons (alpha-helices and beta-sheets) in teal and blue overlaid on a genomic heatmap grid, lab equipment in foreground, dark navy background — AI-generated illustration of biological foundation model research — conceptual reference for the multi-modal molecular data direction.

	Position A (×2)	Position B (×1)
Position type	Postdoc	Postdoc
Focus	Biological foundation models	Frontier generative AI
Biology background	Required	Not required
Stipend	Competitive EPFL postdoc salary + full social benefits	Competitive EPFL postdoc salary + full social benefits
Compute	CSCS / Swiss AI Initiative GPU allocation	CSCS / Swiss AI Initiative GPU allocation
Location	Lausanne, Switzerland	Lausanne, Switzerland
Deadline	Rolling — ideal start: fall 2026	Rolling — ideal start: fall 2026
Apply	bunnelab@gmail.com · `[AIMM Lab] Application PostDoc`	bunnelab@gmail.com · `[AIMM Lab] Application PostDoc`

Continuing open positions

These three positions were listed in prior issues and remain open. No status changes were detected during this week's check.

ISTA — Christoph H. Lampert, CVML group (trustworthy ML / neuro-symbolic AI)

Postdoc, rolling deadline. 7 The CVML group page notes "no fixed deadline, applications will be considered until all positions are filled" and was last updated in June 2026. Research clusters: trustworthy ML (privacy, fairness, robustness, federated learning, LLM security); transfer and continual learning; neuro-symbolic AI. Salary: €73,500–€80,472/year on Austria's collective agreement (14 payments/year). No mandatory teaching. Apply via single PDF to chl(at)ist.ac.at.

ISTA — Francesco Locatello group (causal learning and AI)

Multiple PhD and postdoc positions, rolling. 8 The group page confirms openings are active ("Multiple PhD student and postdoc positions are available!"). Research: causal representation learning, causal discovery, dynamic systems, visual reasoning, AI for Science. The group had 5 papers at NeurIPS 2025. PhD applications go through https://phd.pages.ista.ac.at/; postdoc inquiries to francesco.locatello@ista.ac.at.

University of Bonn — Marc Rußwurm, MEO Lab (geospatial neural fields)

One to two PhD positions, rolling. 9 "The position is open until filled, and applications will be reviewed on a rolling basis." Research: geospatial neural fields — continuous, queryable, uncertainty-aware representations of the planet encoding location, time, scale, sensor modality, and environmental context. Funding: DFG Emmy Noether project + Taylor Geospatial Institute. Salary: TV-L E13, full-time. Target start: late summer 2026. Apply via Tally form at tally.so/r/LZk0Qy (cover letter + CV in English).

Positions that closed this week

All four positions from last week's imminent-deadline cluster have now closed. They remain here for reference only — no applications are possible.

Closed	Advisor(s)	Institution	Sub-field	Status
June 16, 2026	Sebastian Tschiatschek	University of Vienna	IRL, safe/aligned AI	Expired — Nature Careers shows "This job has expired"; Univie portal returns no results 10
June 17, 2026	Robert Jenssen et al.	UiT Arctic AI² Institute, Norway	Deep learning, climate/remote sensing	Removed — Jobbnorge listing withdrawn 11
June 18, 2026	Mathias Verbeke	KU Leuven, Bruges	Causal ML, industrial root cause analysis	Closed — AcademicPositions shows "This job has been closed and is no longer available" 12
June 21, 2026 (stale)	Giovanni Neglia & Chuan Xu	INRIA Sophia Antipolis	Federated learning, privacy/security	Stale — deadline passed, listing page still accessible but no new applications can be considered 13

Complete deadline table

Deadline	Advisor(s)	Institution	Sub-field	Type	First listed
July 17, 2026	Cordelia Schmid	INRIA Paris, WILLOW team	Robotics / Computer Vision	PhD	This issue
September 1, 2026	Melih Kandemir	SDU ADIN Lab, Denmark	Neuromorphic RL	Postdoc	This issue
Rolling	Charlotte Bunne	EPFL AIMM Lab, Switzerland	Biological foundation models	Postdoc ×2	This issue
Rolling	Charlotte Bunne	EPFL AIMM Lab, Switzerland	Frontier generative AI	Postdoc	This issue
Rolling	Christoph H. Lampert	ISTA, Austria	Trustworthy ML, neuro-symbolic AI	Postdoc ×2	June 15
Rolling	Francesco Locatello	ISTA, Austria	Causal learning and AI	PhD & Postdoc	June 15
Rolling	Marc Rußwurm	University of Bonn, Germany	Geospatial neural fields	PhD ×2	June 1
~~June 16~~ ❌	Sebastian Tschiatschek	University of Vienna	IRL, safe/aligned AI	PhD	June 1
~~June 17~~ ❌	Robert Jenssen et al.	UiT [AI]² Arctic Institute	Deep learning, climate/remote sensing	PhD ×5	June 1
~~June 18~~ ❌	Mathias Verbeke	KU Leuven, Bruges	Causal ML, industrial root cause analysis	PhD	May 18
~~June 21~~ ❌ (stale)	Giovanni Neglia & Chuan Xu	INRIA Sophia Antipolis	Federated learning, privacy/security	PhD	June 8

Notes for applicants

INRIA / Schmid closes July 17 — this is the only fixed deadline among the new positions. The ZRR clearance requirement is the one non-research factor that could invalidate an application after the fact; non-French nationals applying should get an explicit confirmation from INRIA's HR team before the deadline rather than after. The research direction fills a gap that has been absent from this roundup since launch — RL/robotics under a world-class CV researcher with a Google dual appointment. 1

SDU / Kandemir closes September 1 — the latest hard deadline in the current pool. The publication bar (≥2 first-author papers at flagship ML venues) is unusually explicit and worth reading carefully before investing time in the application. Applicants who have publications at AISTATS qualify; NeurIPS/ICML/ICLR are the obvious matches. The two-year postdoc contract and Danish social benefits package are competitive for Scandinavia. 3

EPFL / Bunne — three postdoc slots, two different profiles. Position A (×2) requires biological data fluency; Position B requires generative AI methods depth. The Swiss AI Initiative compute allocation (CSCS) is a material advantage for researchers working at large scale. Both positions are rolling but targeting fall 2026 starts, which means earlier applications are more likely to get full consideration. 5 6

ISTA / Lampert and ISTA / Locatello — both groups are at the same institute in Klosterneuburg, Austria. Applicants who fit either direction should consider both simultaneously, as the groups co-affiliate with the ELLIS Unit Vienna. 7 8

NLP/LLM and ML theory gaps persist. No positions in either sub-field qualified this week or last. The INRIA pipeline was particularly thin this cycle — four of the five new INRIA listings screened had PIs without top-venue ML publications. US-based openings remain a structural gap in the covered sources.

Cover image: AI-generated illustration.