Abstract
Data platform sits at the intersection of infrastructure, governance, operations, and user experience. As AI becomes part of everyday engineering workflows, it creates new ways for platform teams to build and operate these systems: summarizing user needs, debugging failures, improving documentation, routing support, generating routine changes, and making self-service workflows easier to use. This talk explores emerging patterns for building data platform with AI while keeping reliability, trust, and governance at the center.
Key Takeaways:
- A practical pattern for AI assisted platform workflows: bringing together context from support threads, docs, logs, runbooks, and ownership metadata, using AI to summarize and propose next steps, and relying on validation checks before anything changes.
- How AI can improve platform operations through incident summaries, runbook suggestions, owner routing, repeated support pattern detection, and follow up tracking.
- How to design AI assisted self service so engineers can use the data platform more easily while access control, policy, auditability, and review stay in the underlying systems.
- Where to draw the boundary between AI suggestions, automated validation, and human approval for critical data platform workflows.
Speaker
Mouli Mukherjee
Engineering Manager @OpenAI
Mouli Mukherjee is an EM at OpenAI for Data Infrastructure. Her team builds the big data backbone at exabyte scale and millions of cores, providing trusted infrastructure for pipelines, queries, and data insights across OpenAI. Its work spans governed data foundations, large-scale pipeline and orchestration systems, and interactive query infrastructure for humans and agents. Mouli is interested in how AI can help build and evolve these systems while keeping reliability, security, and trust at the center.
