10 PhD & Postdoc Openings From Top-Conference ML Labs — Week of May 9–16, 2026

Every entry below comes from a PI who has published at a recent NeurIPS, ICML, ICLR, CVPR, or ACL edition (2024–2026). Positions were publicly posted between May 9 and May 16, 2026. Three additional openings are flagged as borderline and listed separately with caveats — their PIs' most recent confirmed top-conference publications fall just outside the 2024–2026 window or could not be independently verified.

Postdoc openings

Justin Solomon — MIT CSAIL + MIT-IBM Computing Research Lab

Field	Detail
PI	Justin Solomon, Associate Professor of EECS, MIT
Position	Postdoctoral Researcher
Research area	Geometric approaches to safe and robust AI
Start date	As early as Fall 2026
Deadline	Open until filled
How to apply	Email jsolomon@mit.edu — CV + short research description (Solomon requests concise, non-AI-generated messages)

Justin Solomon, Associate Professor at MIT EECS/CSAIL

Image from: Postdoc in Geometric Approaches to Safe and Robust AI at MIT CSAIL

Lab profile. Solomon leads the Geometric Data Processing (GDP) Group at MIT CSAIL, working at the intersection of geometry, optimal transport, and machine learning. The postdoc is jointly hosted with the MIT-IBM Computing Research Lab and targets five specific research threads: data generation, gradient-free optimization, continual learning, the geometry of fine-tuned models, and the geometry of learned embeddings. 1

Recent papers. "Locality in Image Diffusion Models Emerges from Data Statistics" (NeurIPS 2025); "Compress then Serve: Serving Thousands of LoRA Adapters with Little Overhead" (ICML 2025); "WRING Out The Bias" (ICLR 2026); "Score Distillation via Reparametrized DDIM" (NeurIPS 2024). 2

Culture signals. Group size is roughly 10–15 members. No public mentoring philosophy statement, but Solomon is an Associate Editor at top venues and is consistently active at NeurIPS, ICML, ICLR, and CVPR. The joint appointment with MIT-IBM suggests exposure to both academic publishing cycles and applied research constraints.

Gökçin Çınar — University of Michigan, IDEAS Lab

Field	Detail
PI	Gökçin Çınar, Assistant Professor of Aerospace Engineering, UMich
Position	Postdoctoral Researcher (in-person, Ann Arbor, MI)
Research area	Agentic AI for aerospace systems design and optimization
Start date	Available immediately
Deadline	Open until filled
How to apply	Email Prof. Çınar — resume, 1-page statement, 3 references, 2–3 papers; subject line: "Postdoc Application – Agentic AI – [Your Name]"

Lab profile. Çınar's IDEAS Lab (Integrated Design of Efficient Aerospace Systems Laboratory) builds AI-assisted computational workflows that connect aircraft design models, multidisciplinary design analysis and optimization (MDAO) tools, knowledge repositories, simulation codes, and human expert judgment into verifiable end-to-end pipelines. The posted role is specifically aimed at someone who can bridge agentic AI methods with real aerospace engineering toolchains. 3

Recent papers. "A Systematic Framework for Selecting AI Modalities" (NeurIPS 2025); "Visual Agentic AI for Spatial Reasoning with a Dynamic API" (CVPR 2025).

Culture signals. Early-stage lab at the intersection of AI and aerospace — small group, strong cross-disciplinary scope. Çınar's own posting framed it as helping "shape the future of AI-augmented engineering," which signals a lab building its research identity rather than slotting someone into a pre-defined role. Best fit for postdocs comfortable with ambiguity in how formal aerospace problems get modeled.

Yinan Yu — Chalmers University of Technology

Field	Detail
PI	Yinan Yu, Assistant Professor, Dept. of Computer Science and Engineering
Institution	Chalmers University of Technology / University of Gothenburg (joint dept.)
Position	Postdoc, 2-year full-time; 1-year extension possible
Research area	LLM-based decision support for multimodal situation awareness
Start date	June 1, 2026
Deadline	June 1, 2026 ⚠️ Deadline is imminent
How to apply	Apply online via Chalmers portal — CV with publications list and teaching experience, personal statement; English PDFs only

Lab profile. Yu's group works on encoding domain-specific knowledge into ML models, with an emphasis on vision-language models and prompt learning. This postdoc focuses on building AI recommendation engines for real-time emergency response — fusing drone, wearable, and thermal sensor data with 3D digital twin environments to produce explainable priority recommendations. 4

Recent papers. "Domain-Invariant Prompt Learning for Vision-Language Models"; ICML 2025 and ICML 2026 papers; CVPR 2025 publication.

Culture signals. Mid-sized Usable AI research cluster; joint department with University of Gothenburg gives the lab access to two institutional networks. No public group size stated.

Xin Eric Wang — UCSB NLP Group (PhD + Postdoc)

Field	Detail
PI	Xin (Eric) Wang, Assistant Professor of CS; + five co-faculty advisors
Institution	UC Santa Barbara, Department of Computer Science, NLP Group
Positions	(a) PhD admissions open for Fall 2026; (b) Postdoc opening advertised on X, May 2026
Research area	NLP, multimodal learning, embodied AI agents, vision-language reasoning
PhD deadline	Standard UCSB CS admissions cycle; GRE not required; indicate preferred advisors in SOP
Postdoc contact	Contact Xin Eric Wang directly via his X post

Lab profile. UCSB's NLP Group is a multi-faculty operation with six active advisors — Wang, William (Yang) Wang, Xifeng Yan, Shiyu Chang, Wenbo Guo, and Simon Todd (Linguistics). The group had 9 papers accepted at ICLR 2026 and CVPR 2026 combined, spanning 7 main conference papers and 2 Findings papers. Wang also serves as Head of Research at Simular, giving the lab ties to applied AI work. 5

Recent papers. Nine publications at ICLR 2026 + CVPR 2026; Agent S3 released (embodied agent benchmark); extensive record across ACL, EMNLP, NeurIPS, ICML.

Culture signals. Well-documented placement record: PhD graduates have gone to Northeastern University, UC Santa Cruz, University of Waterloo, UT Dallas, Rutgers, Ohio State, and Peking University; industry placements include Google, Meta, Apple, xAI, OpenAI, Amazon, Microsoft, DeepMind, and Salesforce. Faculty attend major conferences (EMNLP, NeurIPS) and encourage students to connect at poster sessions. CSRankings placed UCSB NLP 4th nationally in NLP for 2018–2021. Group funding exceeds $20 million. 5

PhD openings

Suzan Verberne + Gijs Wijnholds — Leiden University, LIACS

Field	Detail
Supervisors	Prof. Dr. Suzan Verberne (primary, NLP) + Dr. Gijs Wijnholds (co-supervisor, formal methods)
Institution	Leiden Institute of Advanced Computer Science (LIACS), Leiden University
Position	Fully funded 4-year PhD candidate (1+3 after positive evaluation)
Research area	Formal methods for NLP — using logical and symbolic methods to constrain LLM fine-tuning and assess generalization
Salary	€3,059–€3,881 gross/month
Deadline	June 26, 2026
How to apply	Online via AcademicTransfer — motivation letter, CV, 1-page research proposal, grade transcripts, writing samples (code samples optional), 2+ academic references

Lab profile. The project sits at the boundary of neural and symbolic NLP. Specific research directions include using logical methods to constrain LLM fine-tuning, combining neural and symbolic approaches for natural language understanding, evaluating LLM generalization in reasoning tasks, studying NLU in the presence of ambiguity, and probing multilingual natural language inference (NLI) in vision-language models. 6

Verberne's recent papers. "A Systematic Survey of Automatic Prompt Optimization Techniques" (EMNLP 2025); "LANCER: LLM Reranking for Nugget Coverage" (ECIR 2026); multiple EMNLP and ACL publications; EMNLP 2024 (Findings). 6

Culture signals. LIACS has ~50+ academic staff. Verberne is an established NLP researcher with reviewer and area chair roles at EMNLP and ACL. Wijnholds contributes formal/compositional semantics expertise. Applicants should have a strong NLP background, solid programming skills, and the ability to submit a credible 1-page research proposal.

Bjoern Andres — TU Dresden, Chair of Machine Learning for Computer Vision

Field	Detail
PI	Prof. Dr. Bjoern Andres, Chair of Machine Learning for Computer Vision, Institute of Artificial Intelligence, TU Dresden
Position	Research Associate / PhD Student (m/f/x), fully funded, salary group E13 TV-L
Research area	Choice of: (a) ML and Image Analysis; (b) ML and Combinatorial Optimization; (c) ML and Automated Theorem Proving
Eligibility	Excellent degree in Mathematics, Computer Science, or Physics; strong mathematical background; excellent English
Deadline	June 1, 2026
How to apply	Apply via DAAD portal

Lab profile. Andres's group focuses on curiosity-driven basic research connecting machine learning with combinatorial optimization and structured prediction — graph decomposition, discrete energy minimization, and lattice problems all appear in recent work. The third track (ML and Automated Theorem Proving) is newer and not widely offered in ML labs. 7

Recent papers. "Box Facets and Cut Facets of Lifted Multicut Polytopes" (ICML 2024); "Partial Optimality in the Linear Ordering Problem" (ICML 2024); "A Sub-Problem Quantum Alternating Operator Ansatz for Correlation Clustering" (ICML 2025). Andres also served on the NeurIPS 2025 and CVPR 2025 Program Committees. 8

Culture signals. Group size not publicly stated. Posting advertises "excellent supervision in a top research environment" and connection to the DRESDEN-concept research alliance, a network of research institutions in Dresden. Strong mathematical orientation — the listing explicitly requires a mathematics, CS, or physics degree background.

Stathis Megas — Medical University of Vienna, MEGA Group

Field	Detail
PI	Dr. Stathis Megas, Accelerate Science Fellow; also affiliated with University of Cambridge
Institution	Center for AI in Medicine (CAIM), Medical University of Vienna
Positions	2 PhD students in AI for Biomedicine, full-time, 4-year contract
Research area	Generative and causal AI for virtual cells, virtual tissues, and virtual whole organs
Start date	September 1, 2026
Deadline	June 3, 2026
How to apply	Submit CV, cover letter, and certificates; quote code number 272/26. Official posting via LinkedIn

Lab profile. The MEGA Group (Mathematical Engineering with Genomics and AI) uses generative AI, causal inference, and mathematics/physics-inspired ML on multimodal biomedical data — single-cell and spatial omics, histology and imaging, and perturbation modeling. Megas framed the lab's mission as building systems capable of modeling how cells, tissues, and whole organs behave and respond to perturbations. 9

Recent papers. ICLR 2025 paper co-authored with Sarah Teichmann (Wellcome Sanger Institute) and Carola-Bibiane Schönlieb (University of Cambridge) — DBLP record: conf/iclr/MegasCPEST25. 10

Culture signals. Newly established group — Megas is actively building his first cohort. Dual affiliation with Cambridge (Accelerate Science Fellowship) and the MetAGE Cluster of Excellence at MedUni Vienna. Co-authorship with Teichmann and Schönlieb indicates access to high-profile collaborative networks. Strong Python/PyTorch skills and a background in machine learning, computational biology, or applied mathematics are listed as requirements.

Borderline openings

The three positions below are included because they are materially relevant — the labs work in core ML/CV/NLP areas and the positions were posted within the monitoring window. However, each has a verification caveat regarding the PI's recent top-conference publication record, noted inline.

Jan Stühmer — HITS gGmbH, Heidelberg (MLI Group)

Field	Detail
PI	Jun.-Prof. Dr. Jan Stühmer, Machine Learning and Artificial Intelligence (MLI) group
Institution	Heidelberg Institute for Theoretical Studies (HITS gGmbH), Heidelberg, Germany
Position	Fully funded PhD (m/f/x), full-time, 3–4 years
Research area	Geometric deep learning for materials science — generative models (diffusion, flow matching) for crystalline materials
Salary	€55,000–€60,000/year
Deadline	August 12, 2026
How to apply	Apply via jobs.ac.uk listing

Caveat. Stühmer's confirmed top-conference publications are from 2023: "Pushing the Limits of Simple Pipelines for Few-Shot Learning" (CVPR 2023) and "Learning Where and When to Reason in Neuro-Symbolic Inference" (ICLR 2023). 11 A 2024–2026 main-conference publication could not be independently confirmed from available records. The HITS MLI group was established in September 2022 and lists NeurIPS/ICML/ICLR as target venues.

Culture signals. Group has roughly 4 PhD students, 3 Master's students, and the group leader. Part of the SIMPLAIX consortium (Heidelberg University + KIT + HITS), backed by the Klaus Tschira Foundation. One of the longer deadlines in this week's roundup — August 12 gives applicants more preparation time.

Mengyan Zhang — University of Bristol

Field	Detail
PI	Dr. Mengyan Zhang, Lecturer, Faculty of Engineering
Institution	University of Bristol, UK
Position	Fully funded 4-year PhD studentship
Research area	Sequential decision-making under uncertainty — reinforcement learning, active learning, Bayesian decision theory, applications in public health epidemiology
Funding	Tax-free stipend at UKRI rate (2025/26: £20,780/year) + RTSG £8,400 + full tuition
Eligibility	UK or international students welcome; competitive worldwide funding
Start date	2027/28 academic year
Deadline	January 31, 2027
How to apply	Online application via University of Bristol; select programme on Programme Choice page and enter studentship details

Caveat. Zhang's ICML publication is from 2021 ("Quantile Bandits for Best Arms Identification"), outside the 2024–2026 window. 12 A NeurIPS 2024 workshop paper was found; a 2024–2026 main-conference paper at the listed venues could not be independently confirmed. That said, Zhang has published in Nature (2025, AI for infectious disease epidemiology) and PNAS, indicating strong scientific output. 13

Culture signals. Early-career PI building a first cohort. Collaborators at Oxford, Imperial College London, and National University of Singapore. The longest timeline in this week's roundup — January 2027 deadline means applicants can begin preparing well in advance.

Mo Lotfollahi — Wellcome Sanger Institute / University of Cambridge

Field	Detail
PI	Dr. Mo Lotfollahi, Group Leader, Wellcome Sanger Institute
Position	Postdoctoral Fellow / Senior Postdoctoral Fellow, 3-year contract
Research area	ML for single-cell and tissue genomics — generative and foundation models for cell fate engineering, perturbation responses, spatial/multi-omics atlases
Methods	Diffusion models, flow matching, transformer-based architectures
Visa	Sponsorship available
How to apply	Application link from LinkedIn post (deadline not stated in post)

Caveat. Lotfollahi's group reports an ICLR 2026 paper (DirMoE) on their lab blog; this could not be cross-verified against ICLR 2026 official accepted papers. 14

Culture signals. Lotfollahi is a Group Leader at the Wellcome Sanger Institute, one of the leading genomics research institutes globally, and is affiliated with the University of Cambridge. The lab has active postdocs and is embedded in the Human Cell Atlas consortium. Co-authored with Sarah Teichmann and other prominent computational biology researchers. Strong fit for ML researchers interested in biology applications who do not require a CS-department institutional home.

Quick reference

PI	Institution	Type	Research area	Deadline
Justin Solomon	MIT CSAIL	Postdoc	Geometric ML, safe/robust AI	Open until filled
Gökçin Çınar	Univ. of Michigan	Postdoc	Agentic AI, aerospace systems	Open until filled
Yinan Yu	Chalmers Univ.	Postdoc	LLMs, multimodal situation awareness	Jun 1, 2026 ⚠️
Xin Eric Wang	UC Santa Barbara	PhD + Postdoc	NLP, multimodal agents	Admissions cycle; contact for postdoc
Suzan Verberne + Wijnholds	Leiden Univ.	PhD	Formal methods for NLP	Jun 26, 2026
Bjoern Andres	TU Dresden	PhD	ML + combinatorial optimization	Jun 1, 2026
Stathis Megas	MedUni Vienna	PhD (×2)	Generative AI for biomedicine	Jun 3, 2026
Jan Stühmer ⚠️	HITS Heidelberg	PhD	Geometric deep learning, materials	Aug 12, 2026
Mengyan Zhang ⚠️	Univ. of Bristol	PhD	Sequential decision-making, public health	Jan 31, 2027
Mo Lotfollahi ⚠️	Wellcome Sanger/Cambridge	Postdoc	ML for single-cell genomics	Not stated

⚠️ = borderline eligibility (see individual entries above)

Applying this week

Cold-email positions (Solomon, Çınar): Both PIs ask you to email directly before or instead of a portal application. Keep the message under 200 words. Solomon explicitly asked for non-AI-generated inquiries, so write yours manually. Attach a CV and a short paragraph on why your work connects specifically to their research threads — not a generic expression of interest.

Portal positions (Verberne/Wijnholds, Andres, Megas, Yu, Stühmer, Zhang): These use institutional HR systems. Most require a research proposal or motivation letter in addition to a CV. For Leiden, a 1-page research proposal is a hard requirement; for Vienna (Megas), remember to quote code 272/26 in your application.

UCSB NLP PhD applicants: GRE scores are not required. Name your preferred faculty advisors in the Statement of Purpose — the group has six faculty with distinct research directions, and specifying who you want to work with matters.

Three imminent deadlines: The Chalmers (Yu), TU Dresden (Andres), and MedUni Vienna (Megas) positions all close by June 3, 2026. If any of these are a fit, treat them as your priority this weekend.