Eventual Blog
What I learned re-evaluating VLADBench on the latest VLMs
GPT-6 Astra and Claude Opus 5.5 top the leaderboard, but the cost-performance frontier is a more interesting story.
Cutting dead frames from DROID: 500 hours of robot data in 32 seconds
The robot state recorded next to the video tells you which frames matter before you decode any pixels. Trimming all of DROID takes 32 seconds on a laptop.
How we made our LeRobot video reader up to 15× faster
LeRobot has emerged as the dominant open format for robot learning data, but decoding frames is expensive. Here's how we made Daft's native LeRobot reader up to 15× faster.
Finding a Needle in the Haystack: Querying Physical AI Data with Daft
Pose + semantic search over Apple's EgoDex hand-manipulation dataset with Daft: SigLIP embeddings meet hand-pose geometry. Ctrl+F for physical AI data.
Disk is the data plane: Flight Shuffle in Daft
How Daft rebuilt distributed shuffle around Arrow Flight, local disk, and streaming reads to handle multi-terabyte workloads.
Scaling As-of Joins
How we built, broke, and re-built our ASOF joins — 5.5x faster, half the memory of pandas, and scaled to a distributed cluster.
Image Embeddings: Tutorial & Examples
Learn about the concept of image embeddings, their various use cases, and best practices for handling them in data processing workflows.
Multimodal Embeddings: Tutorial & Examples
Learn multimodal embedding techniques for cross-modal search, recommendation systems, and content moderation applications.
Tuning Daft's Distributed UDFs: Lessons from ByteDance
Learn from the ByteDance Volcengine LAS Team on how to optimize Daft UDFs on Ray. Discover the formula to evenly distribute data across actors.