[Remote] AI Performance Optimization Engineer

Note: The job is a remote job and is open to candidates in USA. Bright Vision Technologies is a forward-thinking software development company dedicated to building innovative solutions that help businesses automate and optimize their operations. They are seeking an AI Performance Optimization Engineer to focus on maximizing throughput, minimizing latency, and reducing costs across training and inference workloads for large neural network systems. The role involves optimizing AI training and inference pipelines and collaborating with cross-functional teams to deliver well-engineered solutions.ResponsibilitiesProfile and optimize end-to-end AI training and inference pipelines for throughput, latency, and costIdentify and eliminate bottlenecks across data loading, model compute, communication, and memoryImplement and tune quantization, sparsity, and pruning strategies to reduce model footprint and accelerate inferenceOptimize distributed training using tensor parallelism, pipeline parallelism, FSDP, and ZeRO-style shardingTune attention implementations using FlashAttention, paged attention, and related techniquesImplement KV cache optimization, continuous batching, and speculative decoding for LLM servingDrive compiler-level optimizations using Triton, XLA, TorchInductor, or TVM, working with the broader ML framework community to land improvements that translate into measurable end-to-end performance gainsOptimize data pipelines, sharding strategies, and storage access patterns for high-throughput trainingBuild and maintain rigorous benchmark suites and regression frameworks across workloadsCollaborate with ML and platform engineering teams to embed best practices in standard pipelinesDrive cost-efficiency improvements through model architecture, hardware selection, and scheduling strategiesEvaluate new hardware and software offerings, and advise on adoptionDocument performance tuning playbooks and share findings broadly across engineering teamsStay current with AI systems research and translate advances into production improvementsSkillsBachelor's or Master's degree in Computer Science, Computer Engineering, or a related fieldSix or more years of experience in performance engineering, ML systems, or HPCStrong proficiency in Python and C++Hands-on experience optimizing deep learning workloads on modern GPUsDeep understanding of distributed training and inference techniquesExperience with profiling tools across CPU, GPU, and distributed systemsFamiliarity with model compression techniques and their accuracy implicationsStrong grasp of memory hierarchies, communication primitives, and parallelism strategiesExcellent measurement, debugging, and analytical reasoning skillsStrong communication and collaboration skillsExperience optimizing LLM inference at production scaleContributions to vLLM, TensorRT-LLM, DeepSpeed, or similar projectsFamiliarity with custom kernel authoring in Triton or CUTLASSExperience with FinOps for AI workloadsPublications or talks on AI systems performanceBenefitsCompetitive base salary commensurate with experience, plus benefits.Full-time, direct W2 with Bright Vision Technologies (no C2C, no 1099, no third-party).No new H1B sponsorship available. H1B transfers welcomed for qualified candidates.Company OverviewBright Vision Technologies is an information technology company that offers software development, AI, and cybersecurity services. It was founded in 2020, and is headquartered in Bridgewater, New Jersey, USA, with a workforce of 51-200 employees. Its website is https://bvteck.com.