Week of June 2, 2026 — CS top-conference PhD recruitment roundup
Four labs with verified PhD and postdoc openings tied to NeurIPS/ICML/ICLR/CVPR/ACL authors — Rutgers ECE (AI security), NUS Computing (EDA+AI), University of Vienna (ML/NLP), and Capital Normal University Beijing — plus a broader signal from the 2026 MLCommons Rising Stars cohort of 39 upcoming faculty.
This week's roundup covers PhD and postdoc openings tied to researchers with recent top-conference publications (NeurIPS, ICML, ICLR, CVPR, ACL). Four labs have active openings with verifiable deadlines or contact windows. Notes on research direction and, where available, lab culture are included to help you decide whether to reach out.
Opening summaries
| Lab / position | Research direction | Location | Stipend / salary | Apply by |
|---|---|---|---|---|
| Kaiyuan Zhang — Rutgers ECE | Agentic AI security & privacy | New Brunswick, NJ (Fall 2026) | Fully funded (standard US PhD RA rate) | Rolling — form open now |
| Wei Li (Neway) — NUS Computing | EDA, AI-hardware co-design | Singapore (July 2026 start) | NUS standard PhD fellowship | Rolling |
| Benjamin Roth Lab — University of Vienna | ML, NLP, deep learning | Vienna, Austria | €3,714.80/mo (full-time equiv.) | Was: Aug 1, 2025 start — confirm current cohort |
| Capital Normal University — College of AI | Computer vision, NLP, ML systems | Beijing, China | Faculty/research package (TBD) | Ongoing faculty search |
Positions not listed above but worth monitoring: the full MLCommons Rising Stars 2026 cohort (39 researchers) contains at least a dozen incoming faculty who will open labs in 2026–27. See the lab-culture section below.
Position details
1. Kaiyuan Zhang — Rutgers University, ECE (Fall 2026, PhD)
Research direction: Security and privacy foundations for real-world agentic AI systems — making LLM agents secure, private, and trustworthy in practice. Recent work includes BrowseSafe (preventing prompt injection in AI browser agents), CENSOR (gradient inversion defense via Bayesian sampling), and FLIP (provable backdoor defense in federated learning). Papers at IEEE S&P, NDSS, ICLR, USENIX Security.1
Lab culture signals: Zhang's homepage explicitly says experience in agentic AI, LLM post-training, and AI security & privacy is "strongly preferred" — the lab is not a generalist ML group. He also welcomes Rutgers undergrads and MS students into projects, which suggests an open-door culture for early-career people to test the research fit before committing to a PhD.
How to apply: Fill out this form, then optionally email
zhangk@ece.rutgers.edu with a brief follow-up. Rutgers undergrads/grads: use subject line [Research Discussion] – Your Name.Loading content card…
Deadline: Rolling. No cutoff listed; reach out sooner rather than later for Fall 2026 entry.
Fit notes: Best suited for applicants who already have one publication or strong preprint in security/privacy for ML or LLM agent systems. Prior internship at Perplexity, Microsoft Research, or similar is a strong signal to Zhang.
2. Wei Li (Neway) — National University of Singapore, School of Computing (July 2026, PhD/postdoc)
Research direction: End-to-End Autonomous EDA — integrating multimodal LLMs to perceive design context, differentiable optimization for physical design sub-tasks, and hardware testing to anchor AI-driven synthesis in silicon reality. Current affiliation: CMU ECE (PhD candidate), transitioning to NUS in July 2026 as an Assistant Professor.2 Publications: ASP-DAC Best Paper (2021), Apple PhD Fellowship (2022, 2024), Qualcomm Innovation Fellowship (2024); joint patent with NVIDIA for differentiable global routing.
Lab culture signals: Li has a record of translating research into industrial deployment (algorithms shipped in Apple's physical design flows, NVIDIA patent). The lab is likely to suit students comfortable working at the ML/EDA intersection, not purely ML or purely circuits. NUS's School of Computing is consistently ranked in the global top 15 for CS.
How to apply: Wei Li's NUS appointment page is not yet live (as of the writing of this roundup). Monitor https://wadmes.github.io/cv/ for the NUS contact details and open-position announcement, which should appear in June–July 2026. Cold-emailing his current CMU address with your CV and a one-paragraph research statement is reasonable while waiting.
Deadline: Rolling, opening confirmed for the lab's launch in July 2026.
Fit notes: Ideal for students with hardware+software co-design experience, ideally with familiarity in EDA tools (Cadence, Synopsys) or chip tape-out experience. An ML-only background without hardware exposure is a weak fit.
3. Benjamin Roth Lab — University of Vienna, ML & NLP (PhD, Austria)
Research direction: Semi-parametric and retrieval-augmented NLP models; robustness under distribution shift; privacy-preserving and responsible ML. Roth's group sits between systems and theory, publishing at EMNLP, NAACL, and ICLR. This opening was originally listed as starting August 1, 2025; the current PhD cohort status is unconfirmed — treat this as a lab to watch for new openings in the 2026 cycle.3
Stipend: €3,714.80/month full-time equivalent (pre-tax, Austria; base increased with experience). This is notably higher than the typical European PhD stipend.
Lab culture signals: The position was listed as "full-time equivalent" which in the Austrian system means the PhD student is formally employed (not just a stipend recipient), with the social protections that entails. Vienna is a relatively affordable Western European capital.
How to apply: The original listing appears to have closed for the 2025 cohort. Check roth.findable.science or Roth's Vienna faculty page for new openings.
Deadline: Confirm with the lab before applying.
4. Capital Normal University — College of AI (Faculty/Researcher, Beijing)
Research direction: Open faculty and research positions across the college; preference for candidates with strong publication records at NeurIPS, ICML, ICLR, CVPR, ACL, or equivalent venues.4
Position type: This is faculty or research scientist, not a student PhD opening — but included because incoming faculty sometimes recruit their own PhD students immediately. If you are a PhD student looking for an advisor who has NeurIPS/ICML credentials and is based in Beijing, watching for who gets hired here is a reasonable strategy.
Deadline / contact: Ongoing recruitment; apply through the CNU College of AI portal or the chinauniversityjobs.com listing linked above.
Broader signal: MLCommons Rising Stars 2026
MLCommons published its 4th annual Rising Stars cohort on May 19, 2026 — 39 researchers (primarily 3rd–6th-year PhDs and a few postdocs) selected from 175+ applicants, representing 26 institutions across North America, Europe, Asia, and Australia.2
Loading content card…
Why this matters for PhD applicants: several of the 39 are finishing their PhDs and will be on the faculty job market in 2026–27. Opening applications to their labs within the next 3–6 months — before the labs are heavily backlogged — is a practical strategy. Research areas well-represented in the cohort:
- LLM inference systems and efficiency (speculative decoding, hybrid LLMs, distributed training): Rui Pan (Princeton), Muyan Hu (UIUC), Lanxiang Hu (UCSD), Yongjun He (ETH Zürich)
- Hardware-software co-design for ML: Akshat Ramachandran (Georgia Tech), Hyungyo Kim (UIUC), Yuzong Chen (Cornell), Alicia Golden (Harvard)
- Trustworthy and safe AI: Vasisht Duddu (Waterloo), Vaidehi Patil (UNC Chapel Hill), Wenjie Qu (NUS), Yiqiao Jin (Georgia Tech)
- Multimodal & embodied systems: Payal Mohapatra (Northwestern, incoming Arm), Stefany Cruz (UW, robotics)
- Biomedical / healthcare AI: Md Mostafijur Rahman (UT Austin), Hyunji Lee (UNC Chapel Hill)
- ML systems and HPC: Efe Sencan (Boston University), Jae-Won Chung (Michigan), Yuetao Chen (CUHK)
The workshop for this cohort is at AMD HQ in Santa Clara, July 30–31, 2026 — if you're attending, use it as a networking opportunity to get a sense of research fit with individual PIs before the formal application season opens.
Applying to new labs: a practical note
When a professor has just started their lab (or is about to), the application dynamics differ from applying to an established group:
- First-cohort advantage: Early PhD students in a new lab often get more direct mentorship from the PI, since the lab isn't yet running on postdocs and senior students.
- First-cohort risk: Funding pipelines and paper pipelines are not yet established. Ask directly what grants are confirmed for the first year.
- What to ask in outreach emails: (1) Is funding confirmed for Fall 2026 / Spring 2027? (2) How many PhD students are you planning to take? (3) Do you expect students to propose their own projects, or is there a defined research agenda?
None of the four positions above require you to be a US citizen unless explicitly stated (the Rutgers position is open to international applicants; the NUS position is based in Singapore and international by default; Vienna employs international students routinely).
Roundup scope: openings with at least one verifiable source link (faculty page, form, job listing, or news article), for PIs with publications in NeurIPS, ICML, ICLR, CVPR, or ACL from the past two years. Next issue: Monday, June 9, 2026.
Add more perspectives or context around this Post.