Benjamin Irwin

Excited to announce Newton, our new open-source physics engine designed for robot learning! In collaboration with Google DeepMind and Disney Research, we've achieved our first milestone - accelerating MuJoCo by 70-100x on GPUs using NVIDIA Warp. Read more about how Newton here: https://lnkd.in/gpfHSuWM

1,602

12 Comments

Open source is more than just software. For the past 6 years, Google and others have collaborated on OpenTitan, a hardware root of trust chip whose hardware and software have been developed in the open. As a bonus, it supports secure boot with a post-quantum cryptographic algorithm, a first for this type of device. https://lnkd.in/gn8acdgX

160

5 Comments

Since recently joining the Scale AI team as SVP of Engineering, I’ve been inspired by how deeply the team understands what it takes to operationalize AI at enterprise scale. Over the past several years, Scale has helped some of the world’s largest enterprises integrate AI across their most complex workflows. Today marks the next chapter in that journey. We’re excited to open-source Agentex, the agentic infrastructure layer in Scale GenAI Platform, built to enable enterprises to manage secure and reliable enterprise AI. We believe Agentex will become the standard layer for hosting and orchestrating AI agents by enabling developers to build freely while giving enterprises the control and reliability they need for mission-critical systems. Starting today, it’s open-sourced and available to everyone. Learn more about how Agentex powers enterprise AI workflows and try it out yourself: https://lnkd.in/gP7kQsMW

468

40 Comments

There's nothing more exciting than seeing world-class engineers innovate on behalf of customers. Here are three ways Snowflake just got smarter. ✨ Seamless Performance: Powered by Snowflake Optima Optima is our new intelligent optimization engine that automatically accelerates customer workloads. Automatic indexing and query optimizations happen behind the scenes with no knobs or extra effort. As a result, customers are seeing up to 15x faster queries on Gen-2 warehouses with zero additional compute or storage costs. Go here to learn more: https://lnkd.in/gxPBzhQZ 🧠 Empowering SQL Developers Details matter. We're making SQL more expressive, flexible and resilient. MERGE ALL BY NAME simplifies column mapping; updates and inserts happen automatically when names match. CONTINUE handlers let you handle exceptions without breaking flow. And Snowflake Scripting UDFs (now GA) bring full procedural logic directly into SQL. Go here to learn more: https://lnkd.in/gSvnnVew 🚀 Expanded Migration Support: Powered by SnowConvert AI Migrating from legacy systems just got easier. SnowConvert AI now supports more sources than ever including Teradata, Databricks, Oracle, SQL Server, Redshift, Azure Synapse, Google BigQuery, Greenplum, Postgres, Vertica, Hive, IBM DB2 and more. It automatically converts code, procedures and logic into Snowflake-native SQL, helping customers modernize faster and with confidence. Go here to learn more: https://lnkd.in/gX3TqdBT We're focused on making Snowflake faster, smarter and simpler every single day. No tuning. No hacks. Just better engineering under the hood, so builders can focus on building. #Snowflake #AIDataCloud #Optima #Performance #SQL #Migrations

278

4 Comments

🧠 How does an intelligent agent remember what matters most, at the right time? Here’s a sneak peek into a layered memory architecture I've been working on—designed to optimize how agents handle hot, warm, and cold memory using Redis, Qdrant, and PostgreSQL. 🔥🟡❄️ From real-time context to persistent profiles, this design aims to bring speed, semantics, and structure together in one cohesive memory model. And yes—it’s not just conceptual; it’s already in implementation! ⚙️🤖 📌 Stay tuned for my next blog where I’ll break this down and show how to plug it into modern agent frameworks. Exciting stuff ahead! #AI #GenAI #RAG #AgenticRAG #AIAgents

38

This semester, I’ve been assisting the 5th-grade class at Palo Alto Unified School District (where my daughter is a student) in learning Scratch. Interestingly, several kids have expressed that Scratch doesn’t feel as “cool” as Python, which I taught them last year in 4th grade. Their main complaint? Scratch lacks actual code. While exploring alternatives, I came across an online teaching platform: Go+ Builder. It introduces a variant of the Go programming language, allowing students to drag and drop control flows into an editor window—bridging the gap between block-based and text-based coding in an intuitive way. https://lnkd.in/dBHN27YP If you’re involved in teaching programming to young learners, I’d love to hear your thoughts on similar tools! 🚀

66

3 Comments

Inspired by Kaivalya Apte's podcast on modern databases and Chris Riccomini's excellent Substack articles, I decided to get my hands dirty with the building blocks of modern storage engines — LSM Trees, WALs, MemTables, and storage/compute separation. That journey led me to build EchoDB — a simple LSM-based database that writes data to object storage with a single-writer model. Along the way, I documented my learnings in these blog posts 👇 1️⃣ From Heartbeats to Fencing Tokens Leader election + correctness foundations for distributed systems 🔗 https://lnkd.in/gR6nDxn3 2️⃣ Crash-Safe Sequence Numbers: Lessons from LSM Databases Ensuring monotonic, durable numbering for writes (ties back to leader election safety) 🔗 https://lnkd.in/gMXJQa3V 3️⃣ Designing a Write-Ahead Log (WAL) for EchoDB First durability primitive: WAL for crash safety 🔗 https://lnkd.in/gU2-vA3Z 4️⃣ Understanding Memtables: The In-Memory Engine of LSM Trees How writes land in memory before being persisted 🔗 https://lnkd.in/g-d4xCNA 5️⃣ From Memtables to Disk: SSTables, Checkpointing, and Crash Recovery Final piece of the write path: persistence to disk/S3 with recovery guarantees 🔗 https://lnkd.in/geQGYEQm ⚡️ EchoDB is still evolving Bloom filters (to cut down read amplification) are not yet added Currently, only sparse indexing is supported Compaction is a work in progress It’s been a fun ride so far 🚀 🔗 Code is on GitHub : https://lnkd.in/gyXn4Bqz

104

1 Comment

Huge set of releases today across the GitHub Copilot experience. 🎉 Broad availability of Agent Mode, MCP support, leading coding models, Copilot Code Review and Next Edit Suggestions. These features together lead to a profound change to what software development feels like, even from the baseline of folks who were already heavy GitHub Copilot users prior to Agent Mode! And much more still to come... https://lnkd.in/gYeeU7se

84

3 Comments