KTH Royal Institute of Technology
Postdoc in Inference for Large Language Models
- Deadline
- 27 Nov 2026 · 5 days left Closing soon
- Location
- Stockholm, Sweden
- Type
- Postdoc
- Duration
- 24 months · 2 years
- Working language
- English
- Posted
- 22 Nov 2026
- Verified
- 22 Nov 2026
About this opportunity
KTH Royal Institute of Technology in Stockholm has a full-time postdoctoral position in inference for large language models at the School of Electrical Engineering and Computer Science. The post belongs to a five-year Wallenberg Scholar project called "Scalable and adaptive inferencing for democratizing AI", which has a budget of 18 MSEK. The project aims to make serving large language models such as ChatGPT much cheaper and less energy-hungry.
What is offered
- A fixed-term postdoc of up to two years, under KTH's postdoc employment agreement.
- A full-time (100%) position based in Stockholm, Sweden.
- A monthly salary. The posting does not give an amount.
- Close work with Professor Dejan Kostic, Associate Professor Marco Chiesa, other faculty and PhD students.
- The chance to contribute to scientific publications and new grant applications.
- Start date by agreement.
What you will do
- Design, build and evaluate distributed systems and networks for machine learning inference.
- Use machine learning concepts to design agentic frameworks.
- Carry out hands-on experimental systems research.
Who can apply
- You need a PhD, or a foreign degree judged equivalent, by the time the hiring decision is made.
- A strong background in distributed systems, networking, programming and operating systems.
- Very good spoken and written English.
- Very good critical thinking, plus skills in C++, Python, Linux and a scripting language.
- Publications in international journals and top conferences.
Things that count in your favour (meritorious):
- A PhD earned no more than three years before the application deadline.
- Teaching ability, and awareness of diversity and equal treatment, especially gender equality.
- Good collaboration skills combined with the ability to work independently.
- Experience in machine learning, GPU programming or large-scale inference systems.
Personal qualities carry a lot of weight. The role may require a security clearance under Swedish law. If it does, the clearance is a condition of employment.
How to apply
- Apply through KTH's recruitment system using the link on the job posting (reference PA-2026-3247).
- Include a CV that covers your relevant experience and knowledge.
- Include copies of your degree certificates and transcripts. Documents not in English or Swedish must be translated into one of them.
- Include a statement of up to two pages on why you want to do research, your academic interests, and how they connect to your past studies and future goals.
- The deadline is 27 November 2026, at midnight CET/CEST.
- Questions: Dejan Kostic (dmk@kth.se) or Marco Chiesa (mchiesa@kth.se).
Requirements
-
Open to all nationalities
-
Early-career researcher
-
Minimum degree: PhD
Postdoc in Inference for Large Language Models
27 Nov 2026 · 5 days left