ML Research Scientist

About Think

Founded in Riyadh in 2025, Think aims to revolutionize AI infrastructure by integrating hardware and software design to maximize computational efficiency. The company is committed to creating the most efficient computing platform globally, enabling users to operate cutting-edge AI systems independently without waste.

Core Values

• Ownership: Taking full responsibility for outcomes beyond assigned tasks, prioritizing mission over comfort and being resource-conscious.
• Agility: Moving swiftly, iterating rapidly, and committing firmly once decisions are made.
• Impact: Building customer-focused solutions with measurable outcomes.
• Mastery: Diving deep into complex challenges and striving for excellence that delights customers.

Role Overview

The position demands posing critical questions that influence platform behavior, performing empirical research on proprietary hardware by studying model interactions and optimizing the orchestration system accordingly.

Key Responsibilities

• Design and conduct experiments on actual hardware fleets instead of simulated environments.
• Analyze and understand model performance issues such as co-location effects, quantization, adapter serving, and mixed-precision execution.
• Translate experimental outcomes into actionable decisions implementable by the orchestration layer alongside engineering teams.
• Maintain rigorous standards in internal publications, including methods, data validity, and limitations.
• Ensure clear communication about the scope and limits of research findings.
• Keep abreast of current literature to distinguish substantiated results from unverified claims.

Candidate Requirements

• PhD or equivalent experience in machine learning, systems, or a quantitatively-focused field.
• Proven ability to manage full experimental workflows including technical instrumentation.
• Proficiency in Python programming and capability to read and modify model and serving codebases.
• Discipline to identify and state research limitations proactively.
• Beneficial but not mandatory: publications in efficiency, quantization, serving, or machine learning systems.
• Experience with adapter techniques, knowledge distillation, or mixture-of-experts in serving environments.
• Demonstrated history of research translated into production-level results.

What Think Offers

• Opportunity to join one of the first advanced technology firms in the region focusing on foundational tech development internally.
• Significant ownership and influence during early company stages.
• Competitive compensation packages appropriate for a startup environment.
• Close collaboration with a senior, tight-knit team.
• Challenging problems that intersect hardware engineering, system design, and scalable AI.