engblogs

Examines how artificial intelligence and generative models accelerate the discovery and design of therapeutics and antibiotics through MIT collaborations, organ-on-chip testing, and translational pipelines from concept to preclinical candidates.

Antonio Torralba headlines MIT’s 2025 ACM Fellows alongside three alumni, underscoring advances in computer vision, machine learning, and human visual perception, plus leadership in AI research and education.

MIT appoints philosopher Brian Hedden as co-associate dean of the Social and Ethical Responsibilities of Computing (SERC) to advance interdisciplinary ethics in computing and AI within the MIT Schwarzman College of Computing.

Choosing meaningful monitoring metrics: blending DORA metrics with DX-driven developer feedback and Backstage-powered health checks to measure platform value and drive better conversations.

From chasing adoption to delivering real platform value, this article traces John Lewis's journey from simple metrics to multi-stage value measurement - lead times, DORA metrics, and a Technical Health score - driving faster, safer software with paved roads and self-service provisioning.

Concise comparison of Antigravity's all-in-one IDE/agent manager and Gemini CLI's lean terminal CLI with headless execution.

Accelerating marketing ideation by learning from historical campaigns to generate brand-consistent images with generative AI, leveraging embeddings, vector search, AWS Bedrock, and OpenSearch Serverless.

Organizations transitioning from GenAI experiments to agentic, goal-driven systems must align data readiness, governance, and identity-aware access with scalable architecture to build trustworthy, integrated agentic AI capabilities.

Explores Kimi K2 Thinking, Moonshot AI's open-source reasoning model capable of chaining 200-300 tool calls with a 1 trillion-parameter Mixture-of-Experts architecture and production-ready deployment implications.

A case study of refreshing Duolingo's core tabs through craft-led design, balancing consistency and simplicity with a scalable design system, cross-functional collaboration, and rigorous QA to deliver a cohesive learning experience.

VfL Wolfsburg expands ChatGPT into a club-wide AI capability, illustrating a scalable approach to embedding AI across a football organization.

A reinforcement learning-driven universal sequence design framework for Polar codes that scales to code lengths up to 2048 for 6G applications, achieving competitive performance with 5G NR and enabling multi-configuration optimization under physics-informed constraints.