Uploaded December 2025 | Updated November 2026, 3 hours ago
At @TIAA, the goal isn’t “more data”, it’s trusted data that business partners can actually use.
I spoke with Arun Kumar Thirunavukkarasu, who works in the data and AI group and leads a product focused on building a centralized data lake for TIAA’s business teams.
They’ve been on this journey for a little over two years and are using:
Amazon S3 as the storage layer
AWS Glue for building data pipelines – aws.amazon.com/glue
Apache Iceberg as the table format in the data lake
AWS Lake Formation and a serverless approach to manage access and governance – aws.amazon.com/lake-formation
AWS Lambda and other native services to keep things scalable and efficient
Their business problem is that they have highly siloed and inconsistent data across systems, making it hard to get a single view and nearly impossible to make consistent, data-informed decisions.
Outcomes so far include a centralized, consistent, trusted data layer, iImproved data quality and stronger feedback from business stakeholders. They are also seeing clear evidence that teams are increasingly using this trusted data to make better decisions.
Arun’s #1 piece of advice: Start with a solid business requirement.
Don’t chase tech for its own sake, clarify the business problem, design the architecture to solve it, then bring in the tools.
At @TIAA, the goal isn’t “more data”, it’s trusted data that business partners can actually use.
I spoke with Arun Kumar Thirunavukkarasu, who works in the data and AI group and leads a product focused on building a centralized data lake for TIAA’s business teams.
They’ve been on this journey for a little over two years and are using:
Amazon S3 as the storage layer
AWS Glue for building data pipelines – aws.amazon.com/glue
Apache Iceberg as the table format in the data lake
AWS Lake Formation and a serverless approach to manage access and governance – aws.amazon.com/lake-formation
AWS Lambda and other native services to keep things scalable and efficient
Their business problem is that they have highly siloed and inconsistent data across systems, making it hard to get a single view and nearly impossible to make consistent, data-informed decisions.
Outcomes so far include a centralized, consistent, trusted data layer, iImproved data quality and stronger feedback from business stakeholders. They are also seeing clear evidence that teams are increasingly using this trusted data to make better decisions.
Arun’s #1 piece of advice: Start with a solid business requirement.
Don’t chase tech for its own sake, clarify the business problem, design the architecture to solve it, then bring in the tools.










