CS PhD Roundup — September 7, 2026: 1 new Responsible AI opening

The September 1–7, 2026 window produced one verified, application-ready opening: the University of Vienna is hiring a founding cohort of three PhD researchers for Martin Pawelczyk's Responsible Machine Learning Group. The call was posted on September 2 and closes on September 18. 1

Quick triage

Direction: responsible AI — data attribution and curation, privacy-preserving machine learning, machine unlearning, multi-agent LLM safety, and mechanistic interpretability.
Advisor and institution: Martin Pawelczyk, Responsible Machine Learning Group, Faculty of Computer Science, University of Vienna.
Position: three PhD researchers; 30 hours a week. The employment term is three years, beginning with a 1.5-year contract that normally extends after the first-year evaluation; strong progress can extend the appointment to four years. 1
Timing: apply by September 18, 2026. The listing gives an ideal start window of June–October 2026 and separately labels the vacancy start as October 15, 2026; ask the group which date applies to your appointment. 1
First action: prepare a CV, cover letter naming your earliest start date and motivation, thesis or extended abstract, and official Bachelor's and Master's transcripts; then use the University of Vienna application page. 1

Responsible AI

University of Vienna: Martin Pawelczyk

Fit. The group works on a practical slice of responsible AI: data attribution, data curation, privacy and machine unlearning for foundation models; safety in multi-agent LLM systems; efficient training; and mechanistic interpretability. The role suits applicants who want to build and evaluate machine-learning methods for responsible-AI questions. 1

Martin Pawelczyk's recent publication record matches the channel's venue screen. His profile lists Train Once, Answer All: Many Pretraining Experiments for the Cost of One at ICLR 2026, alongside NeurIPS 2025 and ICLR 2025 work on data attribution and unlearning. 2 The ICLR proceedings list Sebastian Bordt and Martin Pawelczyk as the authors of Train Once, Answer All. 3

Role and application. The University of Vienna is recruiting three PhD researchers as the founding cohort of a new group. The work can focus on data-centric AI — including attribution, curation, and privacy — agentic AI, or explainable AI with an emphasis on mechanistic interpretability. The advert asks researchers to develop and publish algorithms, pursue a doctoral thesis, present work at international venues, and contribute to teaching, supervision, and group or faculty administration. 1

Applicants need a completed or near-complete Master's degree in computer science, machine learning, mathematics, physics, statistics, or a related quantitative subject; a foundation in machine learning, statistics, or mathematics; Python; and an early research record such as a strong Master's thesis, open-source contribution, or workshop paper. Experience training large language or vision-language models, distributed training, SLURM, Linux workflows, a research lab, publications, and teaching are advantages. 1

Submit an academic CV, cover letter, thesis or extended abstract, and official transcripts through the application page by September 18, 2026. The advert also links an optional form. Questions can go to Martin Pawelczyk at martin.pawelczyk@univie.ac.at. 1

Lab signal. The advert describes the cohort as early members of a fast-growing team who will help shape the group's research culture and projects. The role also sits in an international academic team and includes access to the Vienna Doctoral School of Computer Science. 1

Deadline watch

University of Vienna — September 18, 2026. Start with the thesis or extended abstract and the cover letter because the application asks for both. State the earliest start date in the letter, then clarify the advert's June–October start window against its October 15 vacancy date if that affects your availability. 1

This week's verified queue has one new opening. Applicants interested in data-centric AI, AI safety, and interpretability can use the September 18 deadline to decide whether a founding-group role at Vienna fits their research record.