AWS is committed to advancing open-source data analytics by contributing upstream to projects such as Apache Spark, Kubeflow Spark Operator, Apache Flink, Apache Iceberg, Apache Airflow, Trino, and Valkey. Learn how we collaborate with open-source communities, build production-ready reference architectures, and share practical lessons from helping customers run these platforms reliably at scale. 

Vara Bonthu, OSS Specialist Solutions Architect, AWS 

Operating a globally distributed, multi-datacenter Apache Cassandra footprint means that minor resource leaks or inefficient configurations can compound into massive infrastructure overhead. In this talk, engineers from Uber's Cassandra team will share how they unlocked immense hardware savings and flattened tail latency. 

Yuqi Yan, Software Engineer, Uber Cassandra Storage Team 
Gopal Mor, Tech Lead Manager, Uber Cassandra Storage Team 

Generative AI has compressed the gap between vulnerability discovery and weaponization from years to days, making fail-open defaults, unsandboxed UDFs, and exposed JMX first-order liabilities. Cassandra ships with the primitives (MutualTLSAuthenticator, role-based authorization, internode TLS, audit logging), but many are off by default; and the database is only the top layer: the JVM, kernel, and OS each carry their own attack surface. This talk maps the most common weakness classes to the controls that close them, and to the missing features (TDE, secrets-manager integration, hardened defaults, stronger UDF isolation) the community should prioritize next. 

Dinesh Joshi, Sr. Engineering Manager, Apple | Apache Cassandra PMC 

Materialized Views (MVs) in Apache Cassandra have historically been avoided by production teams due to reliability gaps and frequent data desynchronization. In this session, we deep-dive into CEP-48, a massive architectural overhaul developed at Uber to bring strong consistency to MVs. We will explore how Uber’s Storage team implemented atomic writes via Paxos v2 (LWT), revolutionized backfills using SSTable streaming, and built an offline Spark validation pipeline to detect and fix inconsistencies. Join us to learn how we transformed MVs from a "disabled feature" into a production-ready system now being contributed back to the Apache Cassandra community. 

Jaydeepkumar Chovatia, Sr. Staff Engineer on Cassandra Team at Uber | Apache Cassandra Committer 
Runtian Liu, Staff Engineer on Cassandra Team at Uber 

Apache Cassandra's roadmap is rapidly accelerating. This presentation highlights major advancements in Cassandra including Transactional Metadata, Trie-Based Memtables and SSTables, Secondary Indexes, and Vector Search; and looks ahead to Cassandra's future in 6.0+: a once-and-for-all replacement for repair, support for distributed ACID transactions, flexible placements, and laying the groundwork for full SQL. 

Scott Andreas, Sr. Engineering Manager, Apple