Mayank Shrivastava·Sep 14The Craft Shifted RightAI agents in a distributed database team: authoring got cheap, and everything after it got expensive.
Yuri Niitsuma·Sep 2Pinot Gateway: closing the observability gap in Apache Pinot with a query logger and an AI feedback…Authors: Yuri Niitsuma with contributions from my team: Tiago Miserani, Vitor Mattioli, Carlos Daniel Fernandes and João Marcelo Alves
Sanjay Thakur·Sep 2Deploying Apache Pinot in Kubernetes: Local Kafka, S3-Compatible Garage Storage, and Zero Cloud…You want to try Apache Pinot. Maybe you’re evaluating it against Druid or ClickHouse. Maybe you need to debug a real-time table config…
Yuri Niitsuma·Aug 28Hallucination-free Text-to-SQL: how we built a data agent with MCP, DataHub, Pinot, and DatabricksHow we turned "how many bets did we take today?" into correct SQL, against the right source, without training a single model.
Shruti Mantri·Jul 23Analyzing Iceberg Tables with StarTreeModern data systems are creating more data every day, and teams need a simple way to store and query it without slowing everything down…
Mohit Ojha·Jul 3The Invisible Transactions: How We Accidentally Broke Apache Pinot’s Upsert Engine (And How We…Imagine this scenario: You are reconciling a massive real-time financial file containing 23.2 million transactions. Your upstream data…
Arunkumar Saravanan·Jul 2Low Pause GC Isn’t Low Latency: G1 vs Generational ZGC on Apache PinotGarbage collection advice for the JVM tends to arrive as folklore. “Use ZGC for low latency.” “G1 is the safe default.” “ParallelGC is…
Gandheharshith·Apr 22Superset Live DashboardFunctional validation on Acceldata ODP (Kerberised Kafka, Pinot, Superset already wired): wiring a realistic payments-shaped stream…
Svetlana Kulish·Apr 15The Real-Time Data Stack Decoded: Where Pinot, RisingWave, and Snowflake FitHello colleagues,
Muralikrishna D·Apr 3The Silent Killer in Real-Time Pipelines: When Flink, Kafka, and Upserts CollideWhen building high-throughput, low-latency data platforms, the architecture usually looks robust on the whiteboard: an Apache Flink…
