engblogs

LOBSTgER combines generative AI and underwater photography to create scientifically accurate and artistically immersive visualizations of marine biodiversity in the rapidly changing Gulf of Maine.

Leveraging AI-driven testing, telemetry analysis, and region-specific simulations to optimize Salesforce Code Builder's performance, scalability, and developer productivity while reducing setup time and infrastructure costs.

AlphaGenome is a cutting-edge AI model that predicts the regulatory effects of genetic variants across long DNA sequences with high resolution, enabling enhanced understanding of genome function and aiding diverse genomic research applications.

Gemini Robotics On-Device is a powerful, efficient VLA model optimized for local execution on robotic devices, enabling fast task adaptation, dexterous manipulation, and robust performance without relying on network connectivity.

Exploring the impact of OpenAI o3, GPT-4.1, and CUA technologies in enabling scalable growth strategies.

Demonstrates how Pinterest leveraged Ray to revolutionize their ML infrastructure by integrating feature development, sampling, and labeling into a scalable, efficient end-to-end ML pipeline, cutting iteration times and infrastructure costs significantly.

Explores Airbnb's approach to enhancing SwiftUI performance through understanding view diffing, leveraging an @Equatable macro for efficient re-rendering, breaking complex views into smaller components, and implementing tooling to manage view complexity at scale.

Gemini Robotics On-Device delivers an advanced, low-latency AI model for local execution on robotic devices, enabling versatile dexterous manipulation and rapid task adaptation without dependency on network connectivity.

Explore essential practices, tools, and challenges in monitoring, evaluating, and optimizing large language models (LLMs) for enterprise applications with a focus on ensuring reliability, safety, and business alignment.

LLMs show sensitivity to nonclinical text variations like typos and gender markers, leading to inconsistent and potentially harmful medical treatment recommendations, especially for female patients.

Salesforce Engineering successfully operationalized AI productivity at scale by integrating AI tools like Cursor, CodeGenie, GitHub Copilot, and Gemini into daily workflows, achieving over 90% adoption across multiple teams and driving measurable improvements in development speed, testing, and knowledge sharing.

An evaluation of local volume measurement using the Tyche library reveals its limited effectiveness for detecting neural network misalignment and anomalous datapoints compared to activation perturbation methods like POSER.