68 articles
Grab transitioned from decentralized to centralized documentation to improve discoverability and consistency at organizational scale.
Cloudflare built Town Lake, a unified data analytics platform, and Skipper, an AI agent, to solve data sprawl across multiple systems.
Databricks operates a large-scale LLM inference platform serving frontier models with 120T+ tokens monthly.
This article explains how Databricks Lakebase architecture is designed to remain resilient to cloud infrastructure failures through high availability principles and strategic system design.
This post addresses CVE-2026-31431 ("Copy Fail"), a security vulnerability in Docker Engine, offering guidance on mitigation strategies.
OpenAI, Thrive, and Crete developed a self-improving tax agent using Codex to automate and optimize tax workflows.
Octopus Energy re-architected its data pipeline to handle Half-Hourly Settlement regulations, achieving a 50x cost reduction.
Databricks enables storing OpenTelemetry AI traces in Unity Catalog as Delta tables, improving observability for production AI agents.
Hugo unifies Grab's data ingestion workflows with Apache Flink, consolidating previously siloed platforms into a single self-service interface.
Databricks Genie is a conversational analytics platform that eliminates the analyst bottleneck by enabling users to ask natural language questions about their data with governed answers.
This article demonstrates how to build a complete real-time fraud detection system using Spark Real-Time Mode and Lakebase, eliminating the need for separate streaming infrastructure.
AI coding agents are now integrated into 60% of developer workflows, enabling autonomous task execution with unprecedented speed, but introducing critical security risks that have led to documented incidents.
Mythos Preview demonstrates significant advances in security-focused AI, but reveals challenges in consistent safeguards and managing vulnerability finding noise.
Grab optimized Android Studio performance in a large monorepo with 2,000+ modules by implementing a custom Focus plugin that reduces IDE sync time.
OpenAI developed a secure sandbox for Codex on Windows to safely run coding agents with controlled resource access.
Cloudflare's billing pipeline slowed after redesigning a ClickHouse table to enable per-namespace data retention.
Superhuman and Databricks partnered to build a 200K QPS inference platform serving grammatical error correction at massive scale.
Databricks built an evaluation framework using LLM judges aligned with human experts through MemAlign to assess the quality of Genie Code-generated machine learning notebooks.
Lakebase achieves 5x faster Postgres writes by eliminating the full page write bottleneck through compute-storage separation.
Grab implemented Shadow Testing in its Flink deployment pipeline to eliminate downtime from production deployment failures.
Grab
Databricks