Case Study: Phind achieves 8x faster AI search responses with Amazon Web Services

Creating a Generative AI Search Engine for Programmers Using NVIDIA-Powered Amazon EC2 Instances with Phind

Phind, an AI search engine for programmers, needed a fast and accurate way to answer complex coding questions at scale. To train and run its LLMs, Phind used Amazon Web Services, including NVIDIA-powered Amazon EC2 P4d and P5 instances, along with AWS ParallelCluster to manage its high-performance computing environment.

Amazon Web Services helped Phind optimize both training and inference speed, with the same infrastructure supporting both workloads. Using AWS and NVIDIA, Phind cut time to first token by 75% and improved tokens per second by 8x, while finding NVIDIA-based EC2 instances to be 2–4 times faster than other options for its workload.

Phind

Michael Royzen

Co-Founder and CEO