Simrat Hanspal is Senior Director of AI Engineering at Applied Data Finance, developing agentic AI for financial services, including portfolio monitoring and identifying delinquency signals. Her specialty is building AI systems that can query business data accurately without exceeding their permissions.
At Hasura in 2023, Hanspal worked in AI engineering and technical evangelism, exploring how vector search and application data could share a governed access layer. She introduced Hasura Notebook, showing how PostgreSQL records could be automatically vectorized into Weaviate and exposed through secured APIs. She also built an AI-powered SRE assistant using ClickHouse infrastructure telemetry and examined the operational requirements of production retrieval, including observability, data freshness, reliability, and authorization.
Her 2023 AI Engineer Summit demonstration crystallized those priorities: an e-commerce assistant generated GraphQL queries combining product descriptions with precise price filters. When unrestricted permissions allowed the model to insert an unintended product, Hanspal replaced them with a read-only role that rejected the same operation. Although the system connected PostgreSQL and Weaviate, the demonstrated searches queried Weaviate alone.
By 2024, Hanspal was a data scientist at consumer-finance company Fi. She subsequently joined Applied Data Finance in a senior AI-engineering role.
- Hybrid retrieval with structured constraints: Combine semantic intent with explicit business rules such as price and category.
- Least-privilege authorization for generated queries: Enforce permissions at the data-access layer so generated searches cannot become unauthorized writes.
- Automatic vectorization of operational data: Synchronize changing relational records with semantic indexes through event-driven embedding workflows.
- Operationally grounded AI assistants: Apply domain-specific telemetry and financial signals to practical investigations, monitoring, and decision support.
