PhD & Postdoc Openings at Top-Conference CS Labs

This Week's Openings (May 11–18)

Seven PhD and postdoc positions caught our attention this week — all from labs that publish regularly at NeurIPS, ICML, ICLR, CVPR, and ACL. Below are the details, grouped by research area.

Postdoctoral AI Researcher — Kempner Institute, Harvard University

Research direction: Foundation models, agentic AI, and AI for scientific discovery. The Kempner Institute wants early-career researchers who can push foundation model training, evaluation, and adaptation forward — especially people interested in agentic workflows, tool-augmented models, and alternative architectures that break from the transformer mold.

Lab culture: The Kempner is a well-funded institute housed within Harvard but operating with startup-lab energy. Postdocs work directly with faculty on high-ambition projects. It is fully in-person — the posting explicitly says remote work is not possible. The $100,000 salary is competitive for Cambridge. Alumni placement includes top industry and academic labs.

Key requirements: PhD within 2 years of start; strong ICML / ICLR / NeurIPS publication record; hands-on PyTorch or JAX experience.

Deadline: July 13, 2026. 1

PhD in Geometric Deep Learning — HITS / MLI Group, Heidelberg

Research direction: Generative models for crystalline materials — using differential geometry and group theory to design improved generative models (diffusion, flow matching) that respect crystal symmetries. Part of the SIMPLAIX consortium linking Heidelberg University, KIT, and HITS. The long-term application targets are battery materials and carbon-capture catalysts.

Lab culture: The MLI group at HITS is an international team with strong theoretical standards. The posting explicitly calls for "high motivation and enthusiasm to work within an international team." PhD students are expected to publish at NeurIPS, ICML, and ICLR.

Key requirements: Master's in CS, math, theoretical physics, or related; strong linear algebra; differential geometry and representation theory are a plus but not required.

Deadline: August 12, 2026 (rolling review). 2

PhD in Mechanistic Interpretability of LLMs — Inria, France

Research direction: Understanding how large language models actually work under the hood — mechanistic interpretability with a focus on problem-solving behavior. The position sits at the intersection of NLP, interpretability, and adversarial ML. Prior exposure to LLMs and NLP is listed as an asset.

Lab culture: Inria is France's national computer science research institute. PhD positions are fully funded with standard French doctoral benefits. Research culture leans toward rigorous, curiosity-driven work rather than product deadlines.

Key requirements: Master's in CS, ML, or related; strong programming; NLP/LLM exposure preferred.

Deadline: Rolling — posted May 13. 3

PhD in Multi-Fidelity Scientific Machine Learning — Inria, France

Research direction: Scientific ML with heterogeneous inputs — building models that combine high-fidelity simulation data with cheaper low-fidelity approximations, applied to physical and engineering systems. Relevant to anyone interested in physics-informed neural networks, surrogate modeling, or uncertainty quantification.

Lab culture: Inria's research culture is curiosity-driven with an emphasis on publication quality over quantity. This project involves close collaboration with domain scientists in physics and engineering, which means you will spend real time understanding the application context — not just tuning models.

Key requirements: Master's in CS, applied math, or engineering; experience with scientific computing and ML frameworks.

Deadline: Rolling — posted May 14. 4

Postdoc in LLM-based Decision Support — Chalmers University, Sweden

Research direction: Building LLM-powered decision-support systems for multimodal situation awareness — think systems that fuse sensor data, text, and domain knowledge to help human operators make decisions under uncertainty. Applications span autonomous systems and emergency response.

Lab culture: Chalmers is one of Scandinavia's strongest engineering schools. Postdocs receive Swedish employment benefits including pension and 5 weeks of vacation. The research group is collaborative and application-driven.

Key requirements: PhD in CS, AI, or related; experience with LLMs and multimodal systems.

Deadline: Recently posted. 5

AI Faculty Positions — Capital Normal University, Beijing

Research direction: Open-rank AI faculty positions in the College of AI. The posting explicitly requires a strong publication record at NeurIPS, ICML, ICLR, and CVPR — a rare explicit conference-list requirement that signals the department's ambition.

Lab culture: Newer AI college within a well-established Beijing university. Chinese academic positions typically come with startup funding and housing support. The explicit top-conference requirement suggests a research-forward culture rather than a teaching-heavy one.

Key requirements: PhD in CS, ECE, statistics, or math; top-venue publications.

Deadline: Recently posted. 6

PhD in ML Efficiency — University of Luxembourg

Research direction: Dynamic sparse training — making deep learning radically cheaper by developing truly sparse solutions for both CPUs and GPUs. The group publishes at NeurIPS, ICML, ICLR, IJCAI, and Nature Communications. Graduated PhDs have landed at Google DeepMind, the ELLIS Institute, Max Planck, UT Austin, and Oxford.

Lab culture: This is a pure-research PhD — the posting explicitly says "no teaching obligations." The group spans three universities (Luxembourg, Eindhoven, Twente) with a hybrid working model. The PI, Decebal Mocanu, runs a high-risk/high-gain lab that encourages summer schools, research visits, and industrial internships.

Key requirements: BSc + MSc in CS, AI, applied math, or related; ambition to publish at A* conferences.

Deadline: Priority deadline was April 16, but applications are processed on receipt — late applications may still be considered. 7

Notes on Using This Roundup

Each week we surface positions where the lab's publication record at top CS conferences is publicly verifiable. If you are applying:

Check the lab's recent papers. A listing that mentions NeurIPS / ICML / ICLR is a signal, but read 2–3 recent papers from the group to gauge whether their research direction genuinely matches yours.
Reach out early. Most of these positions review on a rolling basis. A concise email to the PI — referencing one of their papers and explaining your fit — almost always helps.
Lab culture is hard to read from postings alone. The notes above are based on what the ads say and what is publicly visible. Talk to current or former lab members before committing to a 3–5 year PhD.

Positions are sourced from public postings on Twitter, academic job boards, and university listings. If you are a PI recruiting PhD students or postdocs and publish at these venues, tag your opening with the conference names — it makes discovery easier. Next week's roundup publishes Monday, May 25.

PhD & Postdoc Openings at Top-Conference CS Labs — May 18, 2026