Research Engineer London, England, United Kingdom San Franc

London, England, United Kingdom;
New York, New York, United States;
San Francisco, California, United States;
Seattle, Washington, United States

Lightning AI is the company behind PyTorch Lightning. Founded in 2019, we build an end-to-end platform for developing, training, and deploying AI systems—designed to take ideas from research to production with less friction.

Through our merger with Voltage Park, a neocloud and AI Factory, Lightning AI combines developer-first software with cost-efficient, large-scale compute. Teams get the tools they need for experimentation, training, and production inference, with security, observability, and control built in.

We serve solo researchers, startups, and large enterprises. Lightning AI operates globally with offices in New York City, San Francisco, Seattle, and London, and is backed by Coatue, Index Ventures, Bain Capital Ventures, and First minute.

We are seeking a highly skilled Research Engineer to help optimize training and inference workloads running on Lightning AI infrastructure. This role sits at the intersection of ML systems, AI infrastructure, performance engineering, and practical research. You’ll work across models, inference systems, and platform infrastructure to improve performance, scalability, and reliability for real-world AI workloads.

This is a highly cross-functional role that combines deep technical problem solving with hands‑on implementation. Successful candidates are comfortable working broadly across the stack — from model behavior and inference systems to distributed infrastructure and developer tooling — while collaborating closely with customers and internal engineering teams to solve complex AI performance challenges.

This role is based in one of our hubs (NYC, SF, Seattle, or London), with a minimum of 2 in‑office days per week and occasional team and company offsites.

• Optimize large-scale training and inference workloads across GPUs, accelerators, and distributed systems
• Work directly with customers to analyze workloads, identify bottlenecks, and improve performance, scalability, and reliability of deployed AI systems
• Develop and improve inference pipelines, model serving systems, and performance‑oriented tooling for production AI workloads
• Design and implement profiling, debugging, and observability tools to analyze model execution and guide optimization strategies
• Work across the software stack to ensure performance improvements are accessible through clean APIs, automation, and seamless integration with the Lightning ecosystem
• Partner with hardware vendors and ecosystem partners to support efficient execution across diverse compute backends (NVIDIA, TPU, and emerging accelerators)
• Contribute to open‑source projects through new features, tooling improvements, documentation, and community engagement
• Stay current with advancements in large-scale inference, distributed training, and ML systems optimization

Required Qualifications

• Strong expertise with deep learning frameworks such as Py Torch
• Experience working with large-scale training or inference workloads
• Familiarity with distributed systems and parallelism strategies (data/model/pipeline parallelism, checkpointing, elastic scaling, distributed inference)
• Strong software engineering fundamentals, including designing APIs, building tooling, debugging complex systems, and shipping production‑quality code
• Experience analyzing and improving performance bottlenecks in ML systems, infrastructure, or distributed workloads
• Excellent collaboration and communication skills, including the ability to work cross‑functionally and partner directly with customers or external contributors
• Ability to work comfortably in ambiguous, fast‑moving environments and operate across multiple layers of the stack
• Bachelor’s degree in Computer Science, Engineering, or a related field

Nice‑to‑Haves

• Experience with inference optimization techniques such as quantization, speculative decoding, mixed precision, memory‑efficient training, or throughput/latency optimization
• Experience with technologies such as CUDA, Triton, Tensor
  
  RT, vLLM, SGLang, Dynamo, or…