Sriram Sankar — Founder

Some of my significant projects: LLM Inference (Anyscale) - combining lessons learned from building search engines and other deep infra systems that I've worked on. Built kernels from scratch in C++/CUDA that outperformed existing state of the art systems. Anyscale orchestration framework - a control plane written in Go that can orchestrate stateful Ray (and similar) workloads on a variety of platforms (AWS, GCP, on-prem, Kubernetes, etc.). Forms the core of Anyscale's product offering. Lucene based search engines - Galene (LinkedIn) and Sia (Uber): Existing search engines at these companies on open source frameworks had reached scaling limits. Also customized for specialized product needs - such as searching for moving drivers willing to accept rides. Unicorn Search Engine (Facebook): I was an early engineer on the search team and built the first production Unicorn vertical (for Places/Local Search). Unicorn has since been greatly improved by very smart engineers that worked on it afterwards. Core Systems (Facebook): Worked on Tupperware (renamed Twine), Configerator, and Thrift/ServiceRouter for a couple of years. Bittorrent-based distribution of search indices (LinkedIn): Search indices are large files (many GBs). They are usually generated in one location (e.g., Hadoop cluster) and need to be distributed to multiple serving replicas. Sometimes snapshots generated on one serving replica need to be distributed to all the other serving replicas. Mergeserver (Google): This is an online AB testing framework for Google search that reduced time from diff submission to exposure to real users from a month to a couple of hours. Java Debugger/VM (Metamata): Built a debugger/VM for Java before the JDK had its debugger and profiler interfaces. We built a Java VM in Java that would execute methods that had breakpoints with full introspection capabilities while all other methods could run at full speed on the underlying VM. JavaCC (Sun Microsystems): Built a parser generator for Java (similar to Lex and Yacc) that had over 20000 users in its heyday (late 90’s).

Stackforce AI infers this person is a Backend-heavy Infrastructure Engineer with expertise in scalable search and AI systems.

Location: San Francisco, California, United States

Experience: 33 yrs