Senior Researcher - GPU Algorithms

About Emergence

Emergence AI is a frontier agentic AI lab advancing safe, autonomous AI through its pioneering neuroformal architecture. We are building the next generation of enterprise AI: systems capable of understanding complex knowledge, reasoning across information, executing multi-step workflows, and taking action with the verifiability and reliability required for mission-critical environments.

Our Enterprise Agent Platform provides the foundation for building, deploying, and operating AI systems in production — from data connectivity and knowledge management to workflow orchestration, observability, evaluation, and governance. We partner with organizations in semiconductor manufacturing, life sciences, and other data-intensive industries to solve complex challenges that require trusted AI systems capable of operating in real-world environments.

Our team brings experience from leading AI research labs and technology companies, including IBM Research, The Allen Institute for AI, Amazon, and Broadcom. With headquarters in New York and offices across California, Spain, and India, we bring together diverse expertise and perspectives to tackle some of the most challenging problems in enterprise AI.

THE ROLE

We are looking for a senior researcher to design GPU algorithms that search, evaluate, and check large computational problems in massive parallel, and to connect those results to formal proof. The GPU does the heavy lifting, exploring state spaces, solving differential equations, and evaluating compute graphs across millions of cases, while producing compact certificates that a proof assistant such as Lean checks on the CPU to deliver a final, machine-verified guarantee.

Applications include GPU-accelerated search, numerical simulation, analog trimming, and semiconductor data analysis, but the core skills are general: making massively parallel computation fast, and its results trustworthy.

WHAT YOU WILL DO

• Design and implement high-performance GPU algorithms for search, optimization, and large-scale checking of compute graphs, from algorithm design to kernel tuning and multi-GPU scaling.

• Map irregular workloads such as tree and graph search, case splitting, branch-and-bound, and constraint checking onto massively parallel hardware with strong load balancing, memory locality, and throughput.

• Build GPU solvers for ordinary and partial differential equations and other numerical methods, with rigorous control of floating-point error through techniques such as interval or validated arithmetic.

• Develop certificate-producing pipelines in which untrusted GPU computation emits compact, checkable evidence (witnesses, bounds, case coverage) that a Lean proof verifies on the CPU, keeping the trusted checking step small and tractable.

• Apply these methods to semiconductor use cases, including analog trimming and calibration, yield and excursion analysis, and verification across massive test and process datasets.

• Set technical direction, mentor researchers and engineers, and contribute to publications, patents, and open-source work.

WHAT YOU BRING

• A PhD, or a Master’s with equivalent experience, in Computer Science, Applied Mathematics, Electrical Engineering, or a related field, with several years of research or industry experience in high-performance computing.

• Deep expertise in GPU programming (CUDA, or equivalents such as Triton, HIP, or JAX/XLA) and parallel algorithm design, including memory hierarchies, warp-level primitives, and performance profiling.

• Strong grounding in numerical methods, including differential equation solvers and floating-point error analysis, and experience with large-scale search, optimization, or constraint solving on parallel hardware.

• Working knowledge of formal methods and proof assistants, ideally Lean 4, or a demonstrated ability to pick them up quickly.

• Strong software engineering skills in C++ and Python, with a track record of shipping fast, well-tested systems.

NICE TO HAVE

Experience with certified or validated computation (proof certificates, interval arithmetic, SAT/SMT proof logging), computer-assisted proofs, GPU-accelerated SAT/SMT solving or model checking, EDA tools, analog and mixed-signal design, or semiconductor test and manufacturing data.