Sriram Krishnan

At Benchling, we are building the best in class Cloud Platform for Life Science R&D (see https://www.forbes.com/sites/amyfeldman/2021/04/14/biotech-rd-startup-benchling-hits-4-billion-valuation-as-the-company-starts-laying-the-groundwork-for-an-ipo). We are growing rapidly! We are hiring engineers with full-stack, data infra, growth, and machine learning backgrounds. Check out https://www.benchling.com/careers/#openings, and feel free to reach out to me if you would like to join us!

I was the Head of Engineering for the Machine Learning Platform for Salesforce Einstein - this included teams working on the Data Lake, the Data Pipelines, the Orchestration framework, Machine Learning Serving Infra, Experimentation, and the Compute back-end. I was a first line Manager, Manager of Managers, and Chief Architect at different stages of growth of Einstein Platform. I grew my team from 15 to more than 60, and hired Directors, Architects, and ICs at all levels. My team built the platform that enabled Salesforce to incorporate AI technologies into various product lines to help our customers make better and more informed decisions. We scaled the platform to more than 10B predictions per day, and supported freemium-scale for several GA applications on top of it. More information about our work: • Overall vision: https://www.wired.com/story/inside-salesforces-quest-to-bring-artificial-intelligence-to-everyone/ • Einstein Platform Use Cases: • • Einstein Prediction Builder: https://help.salesforce.com/articleView?id=custom_ai_prediction_builder.htm • • Einstein Case Classification: https://help.salesforce.com/articleView?id=cc_service_what_is.htm • • Einstein Engagement Scoring: https://help.salesforce.com/articleView?id=mc_anb_einstein_engagement_scoring.htm My timeline at Salesforce: • Aug 2019 - Sep 2020: VP of Engineering, Einstein Platform • Oct 2017 - July 2019: Senior Director of Engineering, Einstein Platform

I managed the Data Platform team at Twitter, which included more than 30 engineers and managers. We built the libraries and services that power all analytics at Twitter - processing billions of events per day in real-time, and processing tens of petabytes per day in batch. We also built applications using these tools for reporting core metrics, and creating core data sources. Some highlights of our stack: • Heron, a real-time distributed stream processing engine: https://blog.twitter.com/2015/flying-faster-with-twitter-heron • Summingbird, a framework for Streaming MapReduce: https://blog.twitter.com/2013/streaming-mapreduce-with-summingbird • Tsar, a framework for time-series aggregations: https://blog.twitter.com/2014/tsar-a-timeseries-aggregator • Scalding, a Scala DSL for writing MapReduce jobs: https://github.com/twitter/scalding • Parquet, a columnar file format: http://parquet.apache.org • Data discovery & consumption at Twitter: https://blog.twitter.com/2016/discovery-and-consumption-of-analytics-data-at-twitter My timeline at Twitter: * Sep 2016 - Aug 2017: Senior Engineering Manager, Head of Data Platform * Apr 2015 - Sep 2016: Senior Engineering Manager, Core Data Libraries & Analytics Data Pipeline * Feb 2014 - Mar 2015: Engineering Manager, Core Data Libraries

I was the Tech Lead for the Big Data Platform. I was responsible for the architecture, design and development of a petabyte-scale data warehouse in the cloud (AWS), using open source technologies such as Hadoop, Hive, and Pig. In particular, I was primarily involved in: • End-to-end architecture and development of Genie, which is our Hadoop Platform as a Service, providing REST-ful APIs to run Hadoop, Hive and Pig jobs on the cloud, and to manage multiple Hadoop resources and perform job mediation across them. Genie is open sourced, and available from Netflix OSS GitHub (https://github.com/Netflix/genie). • Building services, tools and templates for complex ETL and analytics patterns in the Hadoop ecosystem. • Managing and fire-fighting petabyte-scale Hadoop clusters on the cloud, which process 100s of TBs of data each day. • Performance benchmarking and monitoring of Hadoop clusters, and pipelines that run on them. • Optimizing performance of Hive/Pig jobs that run on these clusters. More details can be found at: http://techblog.netflix.com/2013/01/hadoop-platform-as-service-in-cloud.html http://techblog.netflix.com/2013/06/genie-is-out-of-bottle.html

• Informatics Lead, Moores Cancer Center Bioinformatics/Biostatistics Shared Resource, UC San Diego (2010 - 2011): Led a team of developers providing advanced data management and processing for research registries and clinical trials, including electronic data capture and validation, and standards-based security. • Technical Lead/Manager, Network for Earthquake Engineering Simulations (NEES) Cyberinfrastructure Center, UC San Diego (2008-2010): Collaborated with earthquake engineers to design a framework for enabling them to remotely participate in experiments, perform hybrid simulations, organize and share data, and collaborate with colleagues, and supervised a team of technical and support personnel involved in the software development process.

• Co-Principal Investigator, “Performance Evaluation of On-Demand Provisioning of Data Intensive Applications”, funded by NSF (2009-2011): R&D in MapReduce and parallel database technology for data-intensive computing, especially in the area of serving and processing of high resolution topographic data sets. Key software developed include myHadoop (http://sourceforge.net/projects/myhadoop), which is a tool for running Hadoop jobs on HPC resources. • SOA Architect, “National Biomedical Computation Resource”, funded by NIH (2004-2010), and the “OpenTopography Facility”, funded by NSF (2010-2011): R&D in the area of “Services-oriented Science”, with a goal to provide Web service based access to scientific applications running on Grid and cluster resources, via workflow and visualization tools. Tools and technologies included Java, SOAP, and cluster management software (SGE, PBS, Condor, DRMAA). Key software developed include Opal (http://opal.nbcr.net), which is a tool for wrapping scientific applications as Web services on grid and cluster resources.

Research Assistant at the Extreme Computing Lab, working on several middleware projects related to Grid computing and Web services technologies. In particular, I was a key contributer to the following projects: • The XCAT (formerly CCAT) framework: XCAT, funded by the US Department of Energy (DOE), is an implementation of the Common Component Architecture specification, which is a software component specification similar to the CORBA 3.0 component specification, but specialized for high performance grid applications. • The XCAT Science Portal, funded by the National Science Foundation (NSF) and the Department of Energy (DOE), which is a Java Servlet based portal implementation that enables scientists to script complex distributed computing applications and package them with simple interfaces for others to use.

• Co-developed Melody, which is a Desktop Grid framework that uses idle cycles on PCs to run computational workloads. A prototype of this system was used to run astronomy simulations on high school PCs in the state of Maryland. • Co-developed Harmony, which is a hypervisor-based approach to building a Desktop Grid for transactional workloads. Requests from clients are scheduled on one of several replicated servers hosted on desktop PCs depending on their observed and predicted loads.

Designed and implemented the Grid Services Flow Language (GSFL), which addresses workflow and composition of Grid services, within the Open Grid Services Architecture (OGSA) framework. This included an XML-based language for GSFL, and a proof-of-concept implementation that demonstrated its features.