Aug 7, 2026 · 34 min
Point Luna builds trustworthy public data pipelines
1535: This Founder Left DoorDash to Organize the World's Data w/ Shyamsunder Sriram
Reliable public data shapes decisions in research, policy, and community life, yet fragmented sources can make even basic analysis misleading.
- 1Point Luna treats missing fields, inconsistent identifiers, and incompatible formats as infrastructure problems, not merely inconveniences.
- 2ChatGPT can locate likely data sources but cannot reliably validate or interpret precise local statistics without trusted underlying systems.
- 3Sriram connects mission-driven company building with alternative funding paths, humility, and a commitment to continuous learning.
Don't miss
Sriram’s diabetes-hotspot example shows why a language model can identify a plausible data source yet still fail at retrieval, validation, or interpretation.
The brief
Sham Sriram, founder and CEO of Point Luna, brings experience from Wayfair, HubSpot, and DoorDash to a problem with broad public consequences: fragmented data makes ambiguous societal questions harder to answer.
Public statistics can mislead when reporting and access to care change, while missing fields, inconsistent identifiers, and incompatible formats make future analysis unreliable even when data exists.
Point Luna uses deterministic pipelines and agentic data engineering to create clean, interoperable datasets; MCP then gives models and less technical users a way to work with validated tables.
Sriram argues that ChatGPT may know where a diabetes statistic lives without retrieving, checking, or interpreting it correctly, exposing the gap between fluent answers and dependable analysis.
The closing turns from infrastructure to temperament: Sriram describes himself as a lifelong wantrepreneur, pairing ambition with humility, openness to being wrong, and continuous learning.
Featuring
Bureau of Labor Statistics
DoorDashBooks & mentions
Listen to the full episode and explore every guest, topic, and moment on PodLume.