Speed Up. Spend Less.™

As a Parallel Computing Runtime Engineer you’ll dive deep into the internals of parallel runtime implementations, design low-overhead interfaces that connect captured execution patterns to our distributed scheduling infrastructure. This role requires strong systems programming skills in C++, comfort working close to the metal, and genuine curiosity about how parallel runtimes work under the hood. Experience with OpenMP, MPI, CUDA, or PyTorch internals is highly valued; the ability and eagerness to learn them quickly is essential.

As a Parallel Applications Specialist, you’ll be the bridge between auto-scaling runtime technology and the real-world workloads it serves. You’ll work with both internal teams and external partners developing Digital Twins and HPC applications—helping them leverage auto-scaling capabilities while identifying opportunities to enhance the platform based on real application requirements. Your insights from the field will directly shape product development priorities.

This role combines hands-on parallel application development with technical consulting and validation work. You’ll benchmark the solution against diverse workloads spanning computational fluid dynamics, finite element analysis, life sciences simulations, and hybrid physics-ML applications. Experience developing parallel versions of HPC or physics-based applications is essential; familiarity with ML techniques for accelerating numerical simulations is a significant plus. You should be comfortable working across the full stack—from application code to runtime behavior—and enjoy both solving technical problems and communicating with diverse stakeholders.

As a Distributed Systems Engineer on the distributed Graph Object Store team, you’ll build the infrastructure that enables the solution to track, move, and accelerate data across distributed computing nodes. The Graph Object Store is the backbone of our distributed execution model—managing object placement, coordinating data movement, and providing the primitives that allow our scheduler to make intelligent placement decisions at runtime.

You’ll design and implement distributed data structures, develop efficient peer-to-peer communication protocols, and optimize memory allocation strategies for high-throughput workloads. This role demands strong knowledge of distributed systems fundamentals: consistency models, distributed key-value storage, memory management, and process/thread coordination. Experience with high-performance networking, RDMA, or distributed storage systems is valuable. You should be passionate about building reliable, performant infrastructure and comfortable reasoning about complex distributed state.

As a Machine Learning Engineer, you’ll develop the intelligent scheduling layer that makes automatic scaling possible. Our scheduling system uses peer-to-peer machine leaning to make real-time load balancing decisions across distributed infrastructure. You’ll own the full ML lifecycle—from model development and training to deployment and online learning in production HPC environments.

Your work will span developing new ML architectures for representing computation graphs, training RL agents that optimize for throughput and resource efficiency, and potentially replacing heuristic components in our distributed object store with learned alternatives. This role offers flexibility to focus on research-oriented model development or production-focused deployment engineering as your interests and the team’s needs evolve. You should have strong foundations in machine learning (particularly GNNs, RL, or both), solid software engineering skills, and enthusiasm for applying ML to systems problems where latency and reliability matter.