Applied Research Engineer

PrimaLabs builds systems that automatically discover the optimal way to run AI workloads on modern hardware. Our platform optimizes model inference across accelerators by intelligently exploring runtime configurations, hardware constraints, and workload characteristics.

At the core of our system is an optimization engine that continuously learns how to improve performance across different models, runtimes, and hardware targets.

Role Overview

PrimaLabs is hiring a Founding Research Scientist to own the intelligence core of our optimization platform.

This role focuses on advancing the scientific foundations of our system: improving surrogate modeling, multi-objective optimization, and transfer learning across hardware targets.

You will extend our optimization stack built on DeepHyper to support new accelerators and inference runtimes, and help establish PrimaLabs as a technical authority through peer-reviewed publications and benchmark results.

This is a scientific moat role your work directly strengthens the core differentiation of the company.

Key Responsibilities

Optimization Research

• Design and improve surrogate models that predict performance across configuration spaces
• Advance multi-objective optimization algorithms for throughput, latency, memory, and cost
• Improve the Pareto frontier quality generated by the optimization engine

Transfer Learning Across Hardware

• Develop methods that transfer optimization knowledge across:
• hardware platforms
• model architectures
• inference runtimes
• Reduce search cost by leveraging historical optimization data

Extending the Optimization Stack

• Extend optimization pipelines built on DeepHyper
• Adapt the system to new inference frameworks such as vLLM and SGLang
• Enable optimization across emerging AI accelerators like:
• NVIDIA H200 Tensor Core GPU
• NVIDIA B200 Tensor Core GPU
• AMD Instinct MI300X

Scientific Credibility

• Publish results in top venues related to:
• AutoML
• ML systems
• high-performance computing
• Develop benchmark studies that demonstrate performance gains on real enterprise workloads
• Collaborate with hardware partners and research groups

Cross-Team Collaboration

• Work closely with ML infrastructure engineers building the optimization engine
• Translate research advances into production systems
• Support customer deployments with scientifically grounded performance improvements

Required Background

• PhD in Machine Learning, Optimization, Computer Science, or a related field
• Research background in:
• AutoML
• Bayesian optimization
• multi-objective optimization
• surrogate modeling
• HPC optimization
• Publications in relevant venues (e.g., ML systems, AutoML, HPC)
• Strong Python and scientific computing experience

Nice to Have

• Experience with hardware-aware ML optimization
• Familiarity with LLM inference systems
• Experience integrating research systems into production ML infrastructure
• Prior work with AutoML frameworks or hyperparameter optimization platforms