Implementing an Enterprise Knowledge Layer: Unlocking Your Data for AI

A knowledge layer between your raw data and your AI/BI tools can connect siloed data and provide crucial business context. Ontologies allow for reasoning, inference, and the ability to query data in a semantically meaningful way. For AI applications, this improves RAG (retrieval augmented generation) and makes output more traceable and explainable. In this talk, I will discuss how I build application-specific ontologies step by step (expanding to a knowledge graph when needed), monitor and evaluate the quality of the ontology using clear metrics, and establish a data flywheel to improve the value of a businesses’ data over time.

Speaker
Pri Oberoi (she/they) is an ML leader who has a track record of improving business metrics at start-ups, in government, and at non-profits over the past decade. Most recently, they provided ML technical leadership at Axios HQ from the ground-up, matured their data strategy as the customer base grew from 0 to >700 companies, and authored their patent. Pri is currently focused on establishing data flywheels and using transparency and interpretability to help users develop better mental models.

Logistics
This event will be in person. We will try to add an online streaming and recording option for the talk if the space, WiFi, and hardware accommodate. If streaming and recording occur, they will be available on the Data Community DC YouTube channel: https://youtube.com/@DataCommunityDC

Agenda
6:30 PM - Food and networking
7:00 PM - Main program starts

After the event, some folks will likely head to Tonic to continue the conversation.

Location
Note that this event will take place in the District of Columbia, two blocks from the Foggy Bottom Metro station at 2112 Pennsylvania Ave NW.