engblogs

This blogpost explores how WebAssembly is revolutionizing web development, enabling faster iteration speed, immediate support across browsers, and improved security, while acknowledging its disadvantages and limitations.

This blog post explores the scope and goals of the recommendation system at Lyft, including customizing app experiences for riders, managing marketplace balance, and recent changes made by the Rider team.

Introducing Koala, a chatbot trained on dialogue data gathered from the web, and its performance compared to ChatGPT and Stanford's Alpaca.

Analyzing TRLX RLHF Transformers using TransformerLens to showcase interpretability of RLHF models

Currency conversion made easy for businesses in the US and Canada using Checkout and Payment Links, allowing dynamic localization of pricing to 30 countries with one click in the Dashboard.

Learn how to deliver enterprise value with practical solutions using foundation models in this blogpost.

Snorkel Flow Spring 2023 introduces new capabilities for adapting foundation models for enterprise use, including warm starts and fine-tuning, along with enhancements for sequence tagging, PDF extraction, embedding import, and the new Studio experience.

Designing optimal neural networks for specific tasks using specialized building blocks to improve accuracy of machine-learning models.

Exploring and iterating on multiple programming models and design strategies to develop a cohesive digital notebook that merges hand-drawn inking with dynamic computing capabilities.

Using machine learning to identify adverse drug events and improve medication prescribing.

A programmable bacterial injection system capable of delivering proteins in mice and human cells, with potential applications in gene and cancer therapies.

Understanding the powerful fraud prevention solution, Stripe Radar, and the key decisions, lessons, and experience gained from building it over the past seven years.