Lastminute.com
## We're excited to welcome two more JVM experts for a morning packed with technical insights. ## Details **Location:** **[Lastminute.com](http://Lastminute.com)** MANTRI COMMERCIO, (Spatium Commercio) TOWER-B, 5th Floor, Marathahalli - Sarjapur Outer Ring Rd, Kariyammana Agrahara, Bellandur, Bengaluru, Karnataka 560103 ***Note: \*\*\* Event from 9.30am to 2pm, Snacks Provided \*\*\**** **\-\- RSVP ONLY IF HAVE YOU DECIDED TO JOIN \-\-** \-\- Its a Free Event\, **RSVP before 27th Aug 2026**. **NO Walk-INs allowed** -- \-\- GOV ID CARD IS MANDATORY TO ENTER \-\- \-\- LAPTOP IS NOT MANDATORY\, BRING IT ONLY IF YOU WISH TO PRACTICE ALONGSIDE \(FULLY CHARGED\)\-\- **Agenda:** ➡️ **Registration Opens @ 9.30 am** ➡️ **Talk #1: The JVM Data Stack Has a Statistics Problem** Modern Java applications increasingly sit on top of massive data systems—search engines, analytical stores, data lakes, and distributed query engines. When those systems make a bad query-planning decision, the JVM application often pays the price, with developers having little visibility into why. A planner is only as good as what it knows about the data. Postgres has spent decades on that: histograms, MCV lists, n-distinct. The JVM data world solved distribution and scale first, which was the right call, and is only now getting to statistics — separately, in every project. OpenSearch is formalizing routing and pruning. Iceberg is designing typed column stats for V4. Parquet carries stats readers barely use. Calcite has the plumbing and no one filling it. The talk opens with one query: fast in Postgres, slow in a search engine, same data. Then it traces where the information gets lost, and covers what a shared substrate might look like — with the upstream work in OpenSearch, Iceberg and Arrow, issue numbers and dead ends included. **Time: 10.00 am to 11.00 am** **Speaker(s) :** [Atri Sharma](https://www.linkedin.com/in/atrisharma) Atri Sharma is a Senior Engineering Manager @ Apple and a techni
Free
Saturday, August 29 · 12:00 AM