| |
|
| |
| DETAILS |
|
IMPORTANT: PLEASE USE THE QR CODE SENT TO YOUR EMAIL TO ACCESS THE BUILDING
Come show your support for your fellow developers!
We're very happy to bring the community together for an evening of learning & connection. This time, we'll have a community member sharing a use case and, as usual, an Elastic employee sharing their expertise as well.
Join us to learn something new & meet others who are passionate about search, observability, & security.
Agenda:
5:30 pm: Doors open; say hi & eat some food
6:00 pm: Agent Builder: Providing Relevant Context for Data Driven Agents", by Kathleen DeRusso, Principal Software Engineer I at Elastic
6:30 pm: Q&A
6:40 pm: Improving Search Ranking with LambdaMART & Elasticsearch, by Brandon (Anbang) Wu, Senior Machine Learning Engineer at Quizlet
7:10 pm: Q&A
7:20 - 7:30 pm: Networking & event wrap-up
Abstract
"Agent Builder: Providing Relevant Context for Data Driven Agents"
What are agents? What is context engineering? Join us as we dive into the importance of context engineering for providing relevant information to agents, & how to build grounded agents using Elastic Agent Builder.
"Improving Search Ranking with LambdaMART & Elasticsearch"
Quizlet's search extended manual Elasticsearch boosts with a Learning-to-Rank (LambdaMART) model. Using query, engagement, & content features, we trained on study signals & validated with A/B tests, achieving measurable lifts in study rates: +3.42% on blended pages, +1.86% on sets pages, & +1.05-1.81% in downstream outcomes. This talk shares how we combined a custom ML re-ranker with Elasticsearch results-covering training data design, offline evaluation, latency, & model interpretability-and offers guidance for Elastic users to go beyond boosting & improve ranking quality.
Where: Elastic NYC Office
1250 Broadway, Floor 16, Training Room
New York, NY 10001
When: Feb 19th | Doors open at 5:30 PM
If you're a long-time Elastic user or just starting your journey, this is the perfect opportunity to share ideas, meet new people, & get inspired. See you there!
|
|
|
|
|
|
|
|
