AI Performance Engineer

About the Role

We are looking for a performance engineer who specializes in making large-scale machine learning workloads fast and cost-efficient in the datacenter. This role is focused on distributed training runs spanning many nodes, and high-throughput batch inference sweeping petabytes of real-world autonomy logs for auto-labeling, data mining, ground-truth generation, and evaluation.

The optimization target here is not tail latency on a vehicle - it is throughput, cluster goodput, and cost per unit of data processed. A training run that wastes 30% of its GPU-hours on stalled data loaders, or an offline inference sweep that takes a week instead of a day, directly slows down how fast the whole company can iterate. You will own the gap between what our fleet of accelerators is theoretically capable of and what our workloads actually achieve: profiling across the stack, finding where the compute and the wall-clock time actually go, and closing the difference.

You will work at the intersection of accelerators, ML frameworks, and large-scale data infrastructure, partnering with the teams who own each layer to land wins that show up in training time-to-result and offline processing cost. At Applied, we encourage all engineers to take ownership over technical and product decisions, closely interact with users to collect feedback, and contribute to a thoughtful, dynamic team culture.

At Applied, you will:

• Profile and optimize distributed training end to end - data loading and preprocessing, augmentation, kernel execution, gradient communication, and checkpointing

• Optimize large-scale offline and batch inference over petabyte-scale sensor logs: batching and scheduling strategies, quantization and low-precision execution, graph optimization, and accelerator saturation across long-running sweeps

• Establish roofline and performance models for our workloads, quantify the gap between achieved and theoretical performance, and stack-rank optimization opportunities by impact and effort

• Improve multi-node scaling efficiency: sharding and parallelism strategies, collective communication, interconnect utilization, and memory-bandwidth and kernel-fusion bottlenecks

• Drive cluster goodput - reduce GPU idle time from input pipeline stalls, storage and network I/O, scheduling gaps, stragglers, and failure recovery on long-running jobs

• Build the benchmarking, observability, and regression-detection tooling that keeps performance from silently degrading as models and code evolve

• Collaborate with engineers across functions to solve complex data and compute problems at scale

• Contribute to a team culture that values effective collaboration, technical excellence, and innovation

We're looking for someone who has:

• Bachelor's degree or higher in computer science or adjacent fields.

• 1 or more years of hands-on experience with ML performance engineering: profiling, roofline analysis, throughput optimization, and root-cause investigation in production systems.

• 5 or more years of overall experience

• Experience with distributed multi-node training at scale (FSDP, DeepSpeed, Megatron, NCCL, or equivalent), including diagnosing scaling inefficiency as node count grows

• Deep familiarity with GPU or accelerator performance concepts - memory bandwidth, kernel launch overhead, occupancy, quantization, collective communication

• Experience with high-throughput or batch inference systems (NVIDIA Triton Inference Server, TensorRT, ONNX Runtime, Ray, or similar)

• Fluency in Python and proficiency in C++ or another systems language

• Excellent debugging, analytical, and problem-solving skills

• A deep understanding of machine learning foundations, and the ability to develop technical solutions for problems with no established playbook

Nice to have:

• GPU kernel development experience: CUDA, Triton, CUTLASS, or hand-tuned attention implementations

• Experience with profiling toolchains such as Nsight Systems/Compute, PyTorch Profiler, or perf

• Experience with GPU scheduling and orchestration on Kubernetes, Slurm, or Ray, including multi-tenant cluster utilization

• Experience with fault tolerance and elastic training for long-running jobs - checkpointing strategy, straggler mitigation, preemption recovery

• Familiarity with autonomy or robotics data (ROS, OpenCV, multi-sensor log formats)