| |
|
|
| |
EVENT DETAILS |
<H2>Overview<BR></H2>
<P>Ever wonder what goes into building & scaling data platforms for learning science, particle physics, rainfall prediction, & cancer research? Come join us for four short talks on<SPAN CLASS="aBn"><SPAN CLASS="aQJ">Wednesday, August 23rd</SPAN></SPAN>at the Quizlet office! These talks are<SPAN>intended to give you a chance to understand the design, implementation, & lessons learned by the engineers who designed data warehousing, pipelining, & management solutions for life science research. We hope you'll learn new things, ask questions, & meet other people at the event.</SPAN></P>
<P><BR></P>
<H2><SPAN>Agenda</SPAN></H2>
<P><SPAN>The event starts with food & drinks at<SPAN CLASS="m_4700464168506196862aQJ"><SPAN CLASS="aBn"><SPAN CLASS="aQJ">6:00 pm</SPAN></SPAN></SPAN>at the Quizlet office,<SPAN>located at 501 2nd St. in SOMA (at the corner of 2nd & Bryant).</SPAN><EM>Please use the entrance on Bryant St. as the back entrance on Federal St. will be closed.</EM><SPAN></SPAN>We'll start the talks at<SPAN CLASS="m_4700464168506196862aQJ"><SPAN CLASS="aBn"><SPAN CLASS="aQJ">6:30 pm</SPAN></SPAN></SPAN>and have additional time for questions, discussion, & refreshments afterwards. Each talk will be 15 minutes.</SPAN></P>
<P><IMG ALT="Headshots of presenters & emcee" SRC="https://cdn.evbuc.com/eventlogos/198496873/headshotbigcopy.png"></P>
<H2>Our Moderator:</H2>
<H3><A HREF="https://www.linkedin.com/in/karenysun/" TARGET="_blank" REL="noopener noreferrer noopener nofollow noopener noreferrer nofollow nofollow noreferrer nofollow">Karen Sun</A><SPAN>(</SPAN><I>Engineering Manager, Quizlet</I><SPAN>)</SPAN></H3>
<P><EM>Karen Sun leads the Backend/API & Data Science teams at Quizlet, focusing on building their data platform to support studying, teaching, & the cognitive science research that powers all the app experiences.</EM></P>
<P><BR></P>
<H2>Speakers & Topics:</H2>
<H3>Optimizing Learning: Cognitive Science & Machine Learning</H3>
<H3><A HREF="https://www.linkedin.com/in/shane-mooney-07250460" TARGET="_blank" REL="noopener noreferrer noopener nofollow noopener noreferrer nofollow nofollow noreferrer nofollow">Shane Mooney</A><SPAN> (Data Scientist - Quizlet)</SPAN></H3>
<P>Every day, millions of students put their learning in Quizlet's hands. It's our job to help them learn as effectively as possible.In order to optimize the study experience & make every question count, we need to present the facts most in need of study. This requires understanding how well each user knows each fact, & tracking how that knowledge state changes over time.<BR> <BR> We'll discuss how, informed by cognitive science research, we used data from the billions of questions answered on Quizlet to train a model that can predict whether a student will get a question correct or not. We use this model to drive question selection in some of the core study experiences on Quizlet while addressing offline support considerations.<BR> <BR> <EM>Shane Mooney was one of Quizlet's first engineers, responsible for building the earliest versions of the Android & iOS apps. He now works on the Quizlet data science team, focused on learning measurement, experimentation, & research.</EM><BR></P>
<H3><BR></H3>
<H3>How We Found the Higgs Boson: Large-scale Data at the Large Hadron Collider</H3>
<H3><A HREF="https://www.linkedin.com/in/brian-calvert-49a373133/" TARGET="_blank" REL="noopener noreferrer noopener nofollow noopener noreferrer nofollow nofollow noreferrer nofollow">Brian Calvert</A> (Data Scientist - Cruise Automation)</H3>
<P>The Large Hadron Collider (LHC), located underground at CERN, produces scientific data at *large-scale* in the first run of the LHC (2011-2012), <I>nearly 100 terabytes</I> of proton-proton collision data were generated <I>each second</I> in the centre of each of the main experiments detectors. This is a necessary price to pay, however, in order to produce the rare fundamental particles of nature that scientists study there. To provide context, only one Higgs boson is produced per O(10 billion) collisions at the LHC. <BR> <BR> Managing these mountains of data requires a comprehensive & powerful data management system, beginning at the data-acquisition stage & following all the way through to the final custom collision databases utilized for statistical particle-physics analyses. This data management is further complicated by the need for a worldwide distributed-data system, so that individual researchers do not have to be onsite to access the data. <BR> <BR> In this discussion, well focus on the data management systems utilized by the CMS experiment, one of the two main experiments at the LHC. We will highlight CMSs main successes in data management & how their design choices greatly facilitate the data analysis of CMSs member physicists. <BR> <BR> <EM>Brian Calvert is a Research/Data Scientist, currently working for Cruise Automation, a company focusing on self-driving cars. Brian spent five years of his research career working on the CMS experiment @CERN, where he was a primary author for multiple peer-reviewed statistical data analyses. Brian also contributed across a broad range of the data management pipeline, including monitoring data-acquisition & worldwide distributed-data processing systems, R&D into event reconstruction algorithms & systems administration for a high-performance scientific computing cluster.</EM></P>
<H3><BR></H3>
<H3>Building a Data Platform for Life Scientists: The Warehouse</H3>
<H3><A HREF="https://www.linkedin.com/in/damon-doucet-9191aa66/" TARGET="_blank" REL="noopener noreferrer noopener nofollow noopener noreferrer nofollow nofollow noreferrer nofollow">Damon Doucet</A> (Engineer - Benchling)</H3>
<P>Benchling is modernizing cancer & other life science research with tools that scientists love to use. Most recently, we've built a data warehouse that aggregates traditionally silo'ed data. The warehouse syncs biological data already stored on Benchling with quantitative results, empowering scientists to perform otherwise impossible analyses & build powerful machine integrations. We'll discuss the engineering challenges involved in syncing updates with often only a few seconds of latency, & surfacing the data to scientists with the rich control over data permissions that the biotech industry requires.<BR><BR><EM>Damon Doucet is an engineer on the data platform team at Benchling, with a background in systems & performance research at MIT.</EM><BR> <BR></P>
<H3>From Research to Production: A New Radar Processing Pipeline for Climate FieldView</H3>
<H3><A HREF="https://www.linkedin.com/in/beth-reid-a3727859" TARGET="_blank" REL="noopener noreferrer noopener nofollow noopener noreferrer nofollow nofollow noreferrer nofollow">Beth Reid</A> (Staff Quantitative Researcher - The Climate Corporation)</H3>
<P>The Climate FieldView platform aims to help the world's farmers sustainably increase their productivity with digital tools. It provides farmers with field-level daily rainfall estimates & real-time rainfall animations using ~500 GB of radar & rain gauge data per day. In this talk I will discuss the science teams rainfall research platform & our work with engineering to implement our algorithms in a production system designed for stability, low latency, & scalability.<BR> <BR><EM>Beth Reid is a Staff Quantitative Researcher at The Climate Corporation. She was previously a Hubble Fellow at Lawrence Berkeley National Lab where she analyzed galaxy maps to measure the properties of gravity on cosmic scales.</EM><BR><BR></P>
|
|
|
|
|
|
