ML Research Scientist

think; Riyadh; On-site; SAR 320,000 - 520,000; Full time

Benefits offered by this job

Ownership and impact

Early-stage compensation

Close collaboration with senior team

Hardware, systems and AI scale

Job summary

think is seeking an ML Research Scientist to run empirical experiments on our own hardware, measuring how models interact at scale and turning findings into product-ready insights. You will design and execute end-to-end experiments, focusing on real fleets rather than proxies, with a strong emphasis on reproducibility and rigorous reporting.

You will work closely with a small, senior team in Riyadh, addressing challenges across hardware, systems, and AI, and contribute to internal publications

Qualifications

• PhD or equivalent research experience in machine learning, systems or a numerate discipline.
• Experience running end-to-end experiments, including instrumentation.
• Strong Python and ability to read/modify model and serving code.
• Ability to state a limitation before others identify it.

Responsibilities

• Design and run experiments on real fleets rather than proxies.
• Investigate model behavior under co-location, quantisation, adapter serving and mixed-precision execution.
• Translate findings into decisions the orchestration layer can act on and collaborate with engineering to implement them.
• Publish internally with rigor: method, data, limits.
• Maintain skepticism about what results support and do not support.
• Track literature and differentiate real results from claims.

Skills

Python

Experiment design

Model serving

Scientific curiosity

Education

PhD-equivalent

Job description

Open roles

ML Research Scientist

Ask the questions whose answers change what the platform does. The work is empirical and it runs on our own hardware: measure what really happens when models share silicon, and turn that into something the product can use.

Type Full time

Team Software

Location Riyadh

Ask the questions whose answers change what the platform does. The work is empirical and it runs on our own hardware: measure what really happens when models share silicon, and turn that into something the product can use.

What you will do

• Design and run experiments on real fleets rather than on synthetic proxies.
• Investigate model behaviour under co-location, quantisation, adapter serving and mixed-precision execution.
• Turn findings into decisions the orchestration layer can act on, and work with engineering to land them.
• Publish internally with the rigour you would use externally: method, data, limits.
• Keep the team honest about what a result does and does not support.
• Track the literature and separate what is real from what is claimed.

What you need

• A doctorate or equivalent research experience in machine learning, systems or a numerate discipline.
• You have run experiments end to end, including the unglamorous instrumentation.
• Strong Python and comfort reading and modifying model and serving code.
• The discipline to state a limitation before someone else finds it.

Useful, not required

• Published work in efficiency, quantisation, serving or systems for machine learning.
• Experience with adapter methods, distillation or mixture-of-experts serving.
• A track record of research that shipped.

What we offer

• One of the first deep tech companies in the region, building foundational technology in house.
• Meaningful ownership and impact at an early stage.
• Competitive early-stage compensation.
• Close collaboration with a small, senior team.
• Problems that combine hardware, systems and AI at scale.