Postgres for AI: Agentic PostgreSQL Lifecycle Management
Presented by:
Ravi Kiran Venkata Bhaskarabhatta
I’m a Senior Solutions Architect – Databases and AI at AWS, working with government, enterprise, and startup customers to design and scale highly resilient, mission-critical data platforms. With deep expertise in PostgreSQL, Oracle, and purpose-built NoSQL databases, I help organizations modernize legacy systems, drive down costs, and adopt cloud-native solutions that power innovation at scale.
I specialize in Amazon Aurora (including Aurora Limitless and DSQL), RDS, DynamoDB, Redis, Cassandra, and have architected globally distributed, ultra-low-latency systems for financial services, public sector, and healthcare industries. I've helped fintech customers build state-of-the-art disaster recovery (DR) architectures with Aurora Global and Limitless, improving P99 latencies by 35%, reducing DR RTO from 3 hours to under 2 minutes, and achieving zero-data-loss resilience with multi-region active-active setups.
With strong experience in big data and analytics platforms, I support customers in building secure, scalable data lakes that fuel GenAI and RAG-based agentic systems. I’ve enabled AI-driven applications for fraud detection, real-time risk modelling, and hyper-personalized experiences, reducing fraud losses by $6M/year and increasing e-commerce MRR by $1.8M.
🎓 Certifications:
AWS Certified Database – Specialty AWS Certified AI Practitioner AWS Certified Solutions Architect – Associate Microsoft Azure SQL DBA Oracle Certified Expert PMP | ITIL
No video of the event yet, sorry!
PostgreSQL has cemented itself as the database of the AI era. With 55.6% developer adoption, major AI platforms scaling to hundreds of millions of users on PostgreSQL, and recent conference keynotes framing PostgreSQL as the foundation for agentic AI workloads, the conversation has shifted from whether PostgreSQL can handle AI-scale demands to how AI can transform the way we manage PostgreSQL itself.
Yet a disconnect persists. DBAs and developers still context-switch between CLIs, consoles, monitoring dashboards, and documentation to perform routine database operations. A single task — diagnosing a slow query, identifying vacuum bloat, applying an index, and verifying the improvement — can span four different tools and twenty minutes of cognitive overhead.
This session introduces agentic PostgreSQL management through the Model Context Protocol (MCP). Using an AI-powered IDE with native MCP server integrations for managed PostgreSQL services, direct PostgreSQL wire protocol connections, and cloud API surfaces, we demonstrate how a single developer can manage the entire PostgreSQL lifecycle from a conversational interface.
The session covers: the Agentic AI + PostgreSQL landscape and three vectors; provisioning and infrastructure as conversation; security and administration without ceremony; deep observability from a chat prompt (pg_stat_*, cloud metrics, Performance Insights); the optimization feedback loop (EXPLAIN ANALYZE → index → verify); and a 10-minute live demonstration against a real managed PostgreSQL cluster.
Target audience: DBAs managing PostgreSQL fleets, developers building on PostgreSQL, platform engineers evaluating agentic AI tooling for database operations.
Takeaways: Ready-to-use SQL queries for vacuum/bloat/slow query monitoring; an architectural pattern combining AI agents with PostgreSQL; clarity on where agentic management adds value vs. introduces risk.
- Date:
- Duration:
- 40 min
- Room:
- Conference:
- PG Down Under 2026
- Language:
- Track:
- Ops and Administration
- Difficulty:
- Medium