Member of Technical Staff (GPU Performance Engineer)

• Design and implement improvements to Reka's training infrastructure and directly contribute to technical decisions that optimise performance of frontier models. • Work on post-training processes including reinforcement learning and fine-tuning. • Contribute to improving the efficiency and scalability of model serving infrastructure as part of a globally distributed foundation model startup.

• Have strong engineering skills with fluency in Python and PyTorch or other frameworks. • Have proven experience implementing and training large deep learning models. • Have experience writing and debugging low-level GPU code in CUDA and C++. • Have experience scaling up GPU jobs using large-scale compute clusters such as Slurm or Kubernetes. • Have demonstrated ability to analyse and optimise the performance of GPU-accelerated workloads, including profiling, identifying bottlenecks, and implementing performance tuning techniques.

• Have experience with post-training processes such as reinforcement learning from human feedback and fine-tuning at scale. • Bring experience optimising model serving infrastructure for latency and throughput. • Have experience with multi-node distributed training across thousands of GPUs. • Have experience with the latest GPU hardware architectures and their performance characteristics.

• Design and implement improvements to training infrastructure that optimise model performance. • Contribute to technical decisions on training efficiency and scalability. • Work on post-training processes including reinforcement learning and fine-tuning. • Improve the efficiency and scalability of model serving infrastructure. • Profile, identify bottlenecks, and implement performance tuning techniques across GPU-accelerated workloads. • Write and debug low-level GPU code to maximise compute utilisation. • Scale GPU jobs across large-scale compute clusters.

• Work fully remotely from the US or UK. • Receive 5 weeks of paid leave. • Receive comprehensive healthcare benefits including vision and dental. • Receive visa assistance including H1B and OPT transfers for US employees. • Train state-of-the-art models leveraging the latest software and hardware. • Collaborate with top-tier engineers and researchers from organisations like Google DeepMind and Facebook AI Research. • Work in an inclusive and open culture that values diverse perspectives.

Reka's mission is to build useful multimodal artificial intelligence and use it to empower organisations and businesses. A globally distributed foundation model startup headquartered in the San Francisco Bay Area, Reka embraces a remote-first approach with top talent from around the world. The founding team and many team members have contributed to many of the breakthroughs in AI over the past decade, with backgrounds at organisations like Google DeepMind and Facebook AI Research.