Compensation Overview

As a Data Scientist you will shape how data is collected, transformed, and used to develop essential new capabilities and features for Yext’s products! We manage large amounts of structured master data on behalf of some of the world’s largest companies and collect transactional data through end-user-facing applications we host directly for these clients. Our mission is to deeply understand user intent, so we can deliver relevant and accurate answers to questions, everywhere people interact with our customer’s digital experiences (mapping services, websites, search, chat, etc.).

Yext clients cover a tremendous breadth of industries. We are seeking data scientists excited to discover opportunities in data across a wide variety of business domains! You will work collaboratively with cross-team partners in software engineering, and product management. You will participate fully in the product development lifecycle and make pragmatic decisions about the design, training, and deployment of machine learning models driving key product features. We are a small, agile and fast paced team, and prioritize high velocity and research progress.

What You’ll Do

• Work tightly with engineers and product managers to ship very cool new ML-powered features in Yext’s products.
• Tackle "full stack" Data Science problems –you will implement solutions to data science problems in a production environment. This means you’ll be great at solving all the challenges that will arise, such as data cleanliness, performance, unexpected feedback loops, and concept drift.
• Data Scientists must help turn raw information into knowledge that enables collaborators to make good decisions. You will articulate Data Science and machine learning concepts to different audiences in an impactful way (e.g. stakeholders in product management, sales, engineering, services, and so on).
• You will investigate technical details (e.g. underlying mathematics, code, or concepts) when called for, but also while demonstrating great judgment and systems thinking about when to simplify, prioritize and abstract when it’s called for.

What You Have

• Enthusiasm for LLMs. Up to date on the latest architectures for all size problems (from ~100M parameter embedding models to 10B+ parameter generative models).
• Experience working with GPU clusters for training and inference of LLMs.
• Expertise in model alignment and evaluation.
• Expertise in Pandas, scikit-learn, matplotlib, PyTorch, CUDA and related tools
• Excellent ability to cope with "glue" problems - shell scripting, debugging python environment problems, workflow automation. Experience with AWS, Linux and Mac environments
• Expertise in databases like Snowflake and MySQL (we use these two), or Presto, Redshift, or other popular systems.
• Knowledge of statistics and good experimental practices (e.g. how to run a powerful A/B test, test for significance, or conduct power analysis for prioritization and planning).
• Knowledge of unsupervised machine learning and information retrieval and mining (e.g. DBSCAN, k-Means, indexing, BM25, and evaluation metrics in these problems).

Bonus Points

• You have an advanced degree in a quantitative field, e.g., computer science, mathematics, business analytics, statistics, applied mathematics, economics, or operations research
• You are autodidactic and self advised – You are able to dive into a new unfamiliar problem domain and start learning about it – You are comfortable with uncertainty and project evolution, and work well on a creative, fast-paced team
• You are an effective communicator: you know how to write a great wiki page, and have phenomenal visual communication skills (an excellent figure is worth a thousand words)
• You are excited to work on training custom LLMs, analyzing data in warehouses, and writing high quality python code.

#LI-JB2