Case Study: Alpha Vertex achieves multi‑GB/s throughput and rapid model retraining with Amazon Web Services (Amazon EFS)

Alpha Vertex - Customer Case Study

Alpha Vertex is a Manhattan-based AI-driven finance firm that builds predictive models (PreCog) to forecast returns for ~20,000 liquid equities. Its model-retraining pipeline ingests millions of data points and runs on a nearly 300-instance Kubernetes cluster that can launch 150 pods per minute and sustain 1,500–2,000 concurrent clients, so the company needed a highly scalable, multi-gigabyte-throughput file-storage solution that could complete retraining within a 48-hour window without slowing production.

Alpha Vertex tested Amazon Elastic File System (Amazon EFS) and deployed it for interim model storage and caching; EFS was 66% faster than traditional NFS and 18% faster than a proprietary pNFS, stood up in under 60 seconds, and scaled to 45 TB within the 48-hour window. Under heavy load it delivered 3–7 GB/s I/O with 1,500–2,000 clients, cost less than the alternatives, and provided reliable, hiccup-free writes with no timeouts or file corruption.

Alpha Vertex

Michael Bishop

CTO and Co-Founder