Senior Distributed Systems Engineer

About the Institute of Foundation Models
The Institute of Foundation Models (IFM) designs and operates ultra-scale GPU supercomputing systems to train next-generation foundation models. We believe performance, fault tolerance, and scalability are co-designed across model architecture, communication systems, runtime, and hardware topology.
This role sits at the core of that effort — driving communication performance, distributed reliability, and cross-layer optimization for large-scale training workloads.

The Mission
We are looking for a deeply technical engineer to co-design and optimize the communication stack for large-scale distributed training, including hybrid parallelism and Mixture-of-Experts (MoE) workloads.
This is not a network operations role. This is a systems-level engineering position focused on performance engineering, distributed debugging, and communication-runtime co-design.
· Design and optimize expert-parallel and hybrid-parallel communication patterns
· Drive high-performance hierarchical collectives for MoE workloads
· Co-design runtime orchestration with communication topology awareness
· Reduce tail latency and improve determinism across thousands of GPUs
· Architect fault-tolerant distributed execution under real-world cluster failures
Core Technical Scope
· Communication-compute overlap and topology-aware collective optimization
· Deep debugging of NCCL, RDMA, and custom communication layers
· Hybrid expert parallel strategies in modern large-scale MoE systems
· Elastic and resilient distributed job orchestration concepts
· Congestion analysis and routing optimization across InfiniBand/RoCE fabrics
· Microbenchmarking and performance modeling for communication-heavy workloads
Expected Technical Depth
· Hybrid expert parallel communication for Mixture-of-Experts training
· Scaling behavior under network pressure
· Distributed orchestration for elastic, large-scale training
· Fault detection and recovery in distributed GPU workloads
· Cross-layer bottlenecks: GPU ↔ NIC ↔ PCIe ↔ NVSwitch ↔ Fabric ↔ Scheduler
Required Background
· Experience optimizing distributed training at 1,000+ GPU scale (or equivalent depth)
· Hands-on expertise with RDMA, InfiniBand, RoCE, and GPUDirect RDMA
· Deep familiarity with NCCL and/or UCX internals
· Strong systems programming ability (C/C++, Rust, or Go)
· Strong familiarity with modern model training frameworks such as PyTorch
· Ability to troubleshoot and profile training performance issues related to communication bottlenecks
· Ability to translate research ideas into production-grade optimizations
· Experience debugging distributed hangs, desynchronization, and performance regressions
What We Mean by "Hardcore"
· You can explain why an communication degrades at scale and how to fix it
· You have improved real cluster throughput via communication redesign
· You can trace a distributed hang across ranks and identify the root cause
· You are comfortable working at the boundary between hardware and runtime
Application Requirements
· Include a link to your GitHub (required)
· Provide links to relevant distributed systems, HPC, or large-scale training projects
· Include a list of publications and/or public technical reports (if applicable)
· Describe the hardest distributed debugging problem you solved
· Include measurable performance improvements you have delivered

Visa Sponsorship
This position is eligible for visa sponsorship.

Benefits Include
*Comprehensive medical, dental, and vision benefits
*Bonus
*401K Plan
*Generous paid time off, sick leave and holidays
*Paid Parental Leave
*Employee Assistance Program
*Life insurance and disability