Agentic Data Repair
3rd place, Lloyds Banking Group hackathon — human-in-the-loop data quality
3rd place at the Lloyds Banking Group internal hackathon (Dec 2025), which opened a few slots to teams from IIIT Hyderabad.
We built an agentic AI system on Google Cloud that detects inconsistencies across a data estate and proposes corrections. The design point was the human in the loop: in a bank, an agent that silently rewrites records is a liability, so every proposed repair is surfaced for a person to approve, reject or amend before anything is committed.