Week of May 12–22, 2026: PhD & Postdoc Openings from Top-Conference CS Authors

Six verified positions from labs with recent NeurIPS / ICML / ICLR / CVPR / ACL papers — each with research direction and what we can tell about the lab from public sources.

Position: PhD students (2027 intake) Location: Chinese University of Hong Kong, Hong Kong Research directions: Embodied AI, Robotic Manipulation, Multimodal Models, Generative Models Stipend: Not disclosed (CUHK PhD studentships typically ~HKD 18,030/month) Application deadline: Not specified — contact directly Apply: hsli@ee.cuhk.edu.hk

Prof. Li was promoted to Full Professor at CUHK in May 2026 1. His group, part of the MMLab network, has a strong publication record across NeurIPS, CVPR, ICLR, and ECCV. The lab recently co-launched Hong Kong's first full-stack embodied AI lab in partnership with 24 industry firms 2, which signals an active pipeline of real-world robotics projects alongside academic publishing.

Lab culture note: MMLab is large and well-resourced, with ties to commercial robotics and the broader AI ecosystem in the Greater Bay Area. Applicants who want industry collaboration alongside top-venue publishing will find the environment well-suited; those seeking a small-lab mentorship style may want to ask directly about group size.

Position: PhD students, visiting students, research assistants Location: Not disclosed publicly (contact to confirm) Research directions: Diffusion Large Language Models (dLLM), Efficient LLM (LoRA, model pruning, model merging), Multi-Objective Deep Learning Stipend: Not disclosed Application deadline: Rolling — contact directly Apply: Via homepage at weiyuchen.cc 3

Chen is actively recruiting across all seniority levels — PhD, visiting student, and RA — which is unusual and means there is a low-barrier entry point for applicants who want to test the collaboration before committing to a PhD program. His recent work spans masked diffusion model sampling paths, multi-task model pruning, and Pareto-front learning for LLM alignment. Publications include NeurIPS 2022 and ICML 2025 work on multi-objective optimization.

Lab culture note: The group's research sits at the intersection of theory (Pareto manifold learning, optimization) and practical LLM efficiency, which suits applicants who want mathematical depth alongside engineering relevance. The open-door policy on RA positions suggests a collaborative onboarding philosophy.

Position: PhD students Location: Philadelphia, PA, USA (GRASP Lab, School of Engineering and Applied Science) Research directions: Robot learning, computer vision, robotics — specifically data-efficient imitation, world-model learning, and robot generalization Stipend: Not disclosed (Penn Engineering PhD typically fully funded) Application deadline: Not specified — formal admissions through UPenn CIS program Apply: Form at PennPAL website for undergrad/MS collaboration; PhD admissions go through the university 4

Jayaraman's lab had two papers at CVPR 2026 and one at ICLR 2026: "Expanding Spatial and Temporal Context for Robotic Imitation Learning Policies With Scene Graphs" (CVPR 2026) and "Correspondence-Driven Trajectory Warping for Data-Efficient Imitation and Autonomous Play" (ICLR 2026). The stated research goal is building general-purpose robots for homes, offices, hospitals, and farms — a broad mandate that translates to varied project types within a unified robot-learning frame 5.

Lab culture note: Jayaraman explicitly encourages women and underrepresented minorities to apply. The lab uses a contact form rather than cold email for prospective students, which keeps outreach structured. He notes difficulty responding to all emails, so the form is the intended channel.

Position: PhD students (ECE or CS program) Location: Madison, WI, USA Research directions: Deep learning architectures, reliable / trustworthy ML (robustness to adversarial attacks, uncertainty, privacy, adaptive algorithms), parsimonious learning (novel architectures, efficient training paradigms) Stipend: Not disclosed (UW-Madison PhD typically funded via RA) Application deadline: Not specified — contact after admission to UW-Madison ECE/CS Apply: grigoris.ece.wisc.edu 6

Chrysos is an Assistant Professor in ECE at UW-Madison. The lab's three stated pillars — machine learning, trustworthy ML, and parsimonious learning — cover both the mathematical foundations (distributional robustness, uncertainty quantification) and the systems angle (new architectures, compute-efficient training). One noteworthy constraint: candidates must already hold an admission offer from UW-Madison ECE or CS before reaching out. This differs from many faculty who accept speculative inquiries; applicants should apply to the program first.

Lab culture note: The lab provides a student prep guide for prospective students currently at UW-Madison, which suggests the PI invests in onboarding. The explicit focus on robustness and privacy is appealing for applicants interested in AI safety as applied research rather than purely theoretical alignment work.

Position: Master's and PhD students (April 2027 intake) Location: Hongo Campus, University of Tokyo, Japan Research directions: Computer vision, vector databases, approximate nearest neighbor (ANN) search, retrieval-augmented generation (RAG), learned data structures Stipend: Not disclosed (UTokyo JSPS/MEXT fellowships available for eligible applicants) Application deadline: UTokyo graduate entrance exam schedule; email first Apply: Email CV + research proposal to Matsui, mentioning the position page 7

Matsui is a Lecturer (senior assistant professor) at UTokyo with recent papers at ICML 2026, CVPR 2026, and SIGMOD 2026. The lab runs on cloud computing (AWS, ABCI, Miyabi), has AI coding environments (Codex), and LLM API access. Core time is none, and the lab explicitly values open-source research release. PhD applicants must have a first-authored paper at a top-tier international conference, or multiple second-tier first-authored papers — the bar is stated upfront and is higher than many assistant-professor labs.

One hard rule: research students are not accepted. If you are not enrolled in UTokyo's graduate program and cannot pass the entrance exam, this is not the route. The lab's niche — large-scale retrieval infrastructure and learned data structures — is less crowded than LLM-centric ML labs, which may appeal to applicants who want to build foundational systems rather than fine-tune models.

Lab culture note: Open-source emphasis and no fixed core hours suggest a flexible, output-oriented environment. The explicit bar on publications before joining implies the group skews toward self-directed researchers.

Position: Postdoctoral researcher Location: Zurich, Switzerland Research directions: Trustworthy ML — mathematical foundations or applied ML reliability: robust distributional generalization, multi-objective learning / alignment, causality, privacy, interpretability Stipend: ETH postdoc salary (Switzerland cost-of-living adjusted; typically CHF 90,000–100,000/year) Application deadline: 2026 start, date flexible Apply: See LinkedIn post for application link 8

Yang is an Assistant Professor at ETH Zurich's CS department and the ETH AI Center, heading the Statistical Machine Learning group. She did her PhD at Berkeley EECS and postdoc at Stanford. Recent work includes theory of LM reasoning and ML reliability at ICML 2025 and ICLR 2025. The posting explicitly looks for candidates who either want to lay mathematical foundations for trustworthy ML, or apply ML reliability methods to real scientific applications — two distinct profiles, making this a broader search than most postdoc listings.

Lab culture note: The posting describes a "supportive, welcoming team" with collaboration across ETH PhD and master's students, plus access to the broader ETH AI community. This is an independent position from the AI Center and ETH postdoctoral fellowships — it is a direct hire into the SML group, which means faster onboarding and no fellowship application dependency.

A note on coverage: This roundup covers openings publicly posted or visible on faculty homepages and professional profiles as of May 22, 2026. Positions are listed based on verified public sources; no paid placements are included. Twitter/X searches returned no results during this collection window — a known platform access limitation. For the most current status of any position, check the linked pages directly before applying.