Fully Funded PhD Positions

ISTA 2027 funded PhD programme: choose a research home through rotations

Decision and advertised opportunity

Strong but verify fit. This is worth prioritising for a strong science applicant who wants substantial research training and several credible group options. It is one institute-wide admission call, rather than a guaranteed opening under a named supervisor. Applicants with either a relevant bachelor's or master's can apply. The original advertisement was placed on 27 October 2026 and describes five-year salaried contracts, conditional continued funding with reasonable progress, and a September 2027 intake. It covers nine advertised subject areas. A candidate attracted solely by a famous professor should first check whether that group appears in the current application portal. Programme and contract overview Current call The application deadline is 8 January 2027, 15:00 CET / 22:00 Beijing. Referee details have the same change deadline; letters are due 12 January 2027, 23:59 CET / 13 January, 06:59 Beijing. The current admissions page specifies a fixed 15 September start. Submit early enough to correct a missing file and monitor references. An application can be incomplete despite having been submitted if its three references do not arrive. 2027 application References FAQ

Research question and doctoral work

The immediate research decision is which two or three plausible groups could become an academic home. The call itself does not prescribe a common experiment, dataset, thesis question or deliverable across astronomy, biology, computing, chemistry/materials, data science, earth science, mathematics, neuroscience and physics. Selecting a track is not equivalent to securing a supervisor. A useful application explains a coherent scientific interest, its connection to several genuinely relevant groups, and evidence that the applicant can contribute. Faculty who are absent from the current portal are not accepting new students in that cycle, although later rotations may still be possible. Current admission rules CS research landscape For example, a computing applicant might compare efficient AI systems with formal learning guarantees. The first asks how memory, communication and numerical choices affect practical training or inference; the second asks which assumptions permit learning a formally specified objective. They share quantitative thinking but demand different daily work: systems implementation and careful measurement versus probability, algorithms and proofs. This comparison is an inference from the three papers below, not an advertised joint project. A physics or life-science applicant must make a corresponding comparison in their own field; these CS examples do not establish fit for every discipline. Alistarh group Chatterjee group A sensible preparation exercise is a small, well-documented analysis or reproduction whose assumptions and limitations the applicant can defend. It is a suggestion, not an admission requirement. Existing code, mathematical coursework, a thesis or an internship can be stronger evidence than a long list of techniques. For an experimental applicant, an equally specific account of their own role and interpretation would serve the same application purpose. There is no evidence here of a required commercial software package, pre-existing publication or guaranteed individual computing allocation. Selection criteria Facilities

University, field and doctoral training

ISTA is a PhD-granting research institute in Klosterneuburg, Austria, rather than a conventional broad undergraduate university. OeAD's institutional entry confirms its research focus and doctoral programme; the official contact address confirms the country. Evaluate it through the relevant groups, training structure and research output rather than a comprehensive university league table alone. OeAD institute entry Official address The advertisement's Nature Index claim can be checked precisely: the 2019 normalized academic table, using 2018 natural-science output, places IST Austria third. Its unnormalized Share position in that same table is 415. The normalization is Share relative to total natural-science article output tracked by Dimensions; it is neither current QS rank nor a ranking of this CS training opportunity. The 2018 Rising Stars statement remains a historical advertisement claim, not an independently rechecked current result. The recent papers and programme rules give more direct evidence for today's application decision. Nature Index methodology and table Students start unaffiliated and undertake three approximately two-month rotations in different groups. Both routes require affiliation by 30 June of year one. Normally bachelor's entrants take the long route, with 42 ECTS coursework and a qualifying-exam deadline of 15 September at the beginning of year three; master's entrants take the short route, with 30 ECTS and a 15 February year-two deadline. These are route deadlines, not promises about every student's actual examination date. A thesis committee includes the supervisor and two other members, including an external member. Later training includes twice-yearly progress reviews and at least one half-semester of teaching assistance. The institute awards the PhD. Programme structure Three- and four-year bachelor's degrees are eligible in principle. The optional MS en route is an additional award with extra curricular requirements, not a necessary intermediate degree or a separately admitted master's programme. The FAQ states that the PhD diploma does not list a subject specialization, which may matter when explaining the degree to future employers or credential assessors. The current programme's shared facilities include scientific computing and multiple experimental service units; actual access and training must be discussed with the future group. Bachelor entrants Degree FAQ Facilities

Supervisors and three recent contributions

There is no assigned supervisor at application. Two illustrative current CS professors are Dan Alistarh, professor since 2022, whose group works on efficient large-model training/inference and distributed computation, and Krishnendu Chatterjee, professor since 2014, whose group includes quantitative verification and stochastic games. Their websites establish research activity, not current portal recruitment, supervision capacity or a reserved place. Individual mentoring style and time availability remain questions for interviews and rotations. Alistarh profile Chatterjee profile LLMQ, by Erik Schultheis and Dan Alistarh, appears in the 2026 PMLR proceedings of CPAL (the meeting is labelled CPAL 2025). The two-author original paper addresses commodity-GPU memory and communication constraints through a lower-precision training implementation, benchmarks and end-to-end tests. The relevant applicant signal is systems work grounded in measurement and numerical tradeoffs. Its limitations section cautions about extrapolating low-precision success to much longer training; it does not establish that every affordable machine can cheaply train a competitive general model. Alistarh is second of two authors; the read sections do not allocate individual technical tasks. Reading covered the complete institutional abstract and selected paper sections through limitations, rather than every appendix. LLMQ record and paper Speculative decoding speed-of-light, by Sergei Pankratov and Dan Alistarh, is an EACL 2026 original conference paper. Its complete institutional abstract reports tight runtime lower bounds for deterministic speculative generation under stated assumptions, using branching random walks and empirical Llama evaluations. It illustrates that faster inference also raises mathematical limits, not merely engineering benchmarks. The result should not be generalized to all stochastic algorithms, hardware or deployment conditions. Alistarh is second of two authors; task-level contribution is unverified. The full paper was not read, so no proof-level assessment is claimed. EACL paper record PAC learning in turn-based stochastic games with reachability objectives, by Ali Asadi, Krishnendu Chatterjee and Pavol Kebis, is an original CONCUR 2026 paper. The abstract and paper front matter describe decentralized learning with private information and an expected conditional distance parameter, with a sample-complexity guarantee in a finite-state reachability setting. Its value for preparation is showing how positive guarantees depend on explicitly modelled assumptions. It does not solve arbitrary adversarial reinforcement learning. Chatterjee is second of three authors; no individual role is inferred from position alone. The proofs were not audited. These three papers are important illustrative outputs from two potential groups, not a review of all ISTA supervisors or the thesis work that will be assigned to this cohort. CONCUR paper record

Funding and living costs

Doctoral students are employees with social security, not recipients of a UK-style tax-free stipend. The official funding page lists 2026 gross monthly salary of €3,053–€3,673, paid 14 times per year, depending on entry qualification and progression through the qualifying exam. This corresponds arithmetically to €42,742–€51,422 gross annually at those reference rates. It does not establish a 2027 salary, net take-home pay, or that every new bachelor's entrant starts at the maximum. There are no tuition or application fees. Conference/workshop support is available, but its individual allowance is not stated. Funding rules OeAD independently confirms five-year employment contracts. The recruitment notice further states that students making reasonable progress are funded through defense. The read official funding page confirms employment and salary, but does not establish an unlimited extension entitlement; obtain progress, extension and group-change conditions in writing if a thesis outlasts the initial contract. The FAQ says ISTA cannot fund visa and relocation expenses. Budget separately for these, a housing deposit and initial travel. OeAD contract overview Employment and funding FAQ OeAD's national student guide gives an approximate €1,300 monthly average, based on the 2025 student survey/IHS 2026, with accommodation around €450–700, food €300 and personal/study spending €500. The component range and total are broad estimates, not a precise additive household budget. Vienna can cost more. Campus apartments are subject to availability; no current individual rent, utilities or deposit was verified. Gross salary minus this national average is not a defensible net-surplus calculation. Request a salary illustration and written housing quote. Living-cost guide Housing

Applicant fit and preparation

The strongest fit is an applicant with relevant scientific foundations, a clear account of their own research contribution, credible options across groups, and willingness to take courses and rotations. Direct bachelor's admission is a real programme feature; a master's is not universally required. International applicants have no separate nationality quota in the FAQ. None of this guarantees individual Chinese degree equivalence, admission or a particular immigration title. There is no verified China-specific percentage cutoff here. Final-year applicants must complete their stated studies before 15 September; the FAQ requires evidence of first-degree completion by August at the latest. Bachelor eligibility Application FAQ The programme expects strong English and assesses it in interview; its FAQ says formal IELTS/TOEFL/GRE scores are not required. There is no fixed general GPA threshold. The FAQ says successful candidates are typically in their institution/field's top quartile and expects top 15% or better from high-volume application origins; this is not a verified China-specific percentage conversion. For preparation, prioritise evidence of research readiness and an interview explanation of choices and limitations, rather than importing a British IELTS profile or inventing an admissions probability. Admissions FAQ Prepare a CV, a SoP normally limited to one or two pages (the FAQ says slight excess alone is not penalised), all years of all degree transcripts/diplomas in original language and English, and three suitable referees. The current 2027 page adds short answers of at most 5–8 lines on up to three groups, an ISTA professor's paper with DOI and how one could contribute, and the applicant's most meaningful research experience. Its annual “up to three” wording differs from the generic FAQ's “three to five”; follow the actual 2027 portal fields and seek clarification if they disagree. Referees submit confidential letters themselves; at least two should come from an academic/research setting. Current materials Reference rules The 18 November 2026 virtual open day offers a useful way to ask about rotations and group availability before committing substantial application time. The general session is 13:00–14:00 CET, followed by concurrent tracks through 17:00. Those times correspond to 20:00–21:00 and later evening in Beijing. Participation is a preparation option, not a stated admission condition. Open day

Risks and questions before accepting

Ask admissions which groups recruit in this cycle, how unsuccessful affiliation is managed, and whether two or three scientifically sensible alternatives exist for your background. Ask prospective supervisors about their current research priorities, feedback routines, student autonomy, shared supervision and how resources are allocated. Confirm the route and examination deadline assigned to your qualification. The rotation structure provides opportunities to investigate fit; it is not a guarantee of joining the first-choice group. Structure Student support Ask HR for the 2027 salary scale, a net-pay illustration, contract/extension conditions, housing availability, and the correct entry/residence/employment procedure. OeAD's researcher residence route requires suitable qualifications, a hosting agreement and other conditions; admission alone does not resolve applicability, especially for bachelor's entrants. ISTA's International Officer and HR explicitly advise on these issues. These are matters for individual confirmation, not a promise of residence approval. OeAD researcher route ISTA HR support Career prospects depend on the eventual project: efficient-computing work can build systems and measurement skills; formal methods can build probabilistic modelling and proof skills; other tracks have different paths. This is a skills-based inference from the research, not measured employment demand. ISTA provides careers training, industry visits, an alumni network and commercialisation support; no verified placement rate, current job count, future salary or visa sponsorship guarantee is available. Prioritise this programme if the research and rotation model fit, and decide on an offer only after the group, employment and move are concrete. Careers and alumni Research facilities and industry interface

Evidence and Sources

• ISTA 2027申请与录取
• ISTA 博士雇佣与经费
• ISTA 博士培养结构
• 本科直博及可选硕士学位
• ISTA 博士申请常见问题
• ISTA 科研设施
• ISTA 职业发展与校友
• ISTA 住宿与通勤区域
• ISTA 2026线上开放日
• ISTA 计算机科学研究组
• Dan Alistarh 官方研究组
• Krishnendu Chatterjee 官方研究组
• LLMQ: Efficient lower-precision LLM training for consumer GPUs
• Speculative decoding speed-of-light: Optimal lower bounds via branching random walks
• PAC learning in turn-based stochastic games with reachability objectives
• 奥地利官方学生生活费参考
• 奥地利研究人员居留许可
• ISTA 新生行政与学术支持
• ISTA 官方联系地址
• OeAD 对ISTA的机构与培养介绍
• Nature Index 2019 标准化学术机构表