Arpit Roopchandani

Primary founding engineer building Patched's production AI platform and banking deployments across customer operations, fraud and risk, and reconciliation.

• Built a platform for scenario-scoped AI workflows with verification, human escalation, and auditable execution; it runs 500,000+ workflows monthly.
• Built Banking Customer Ops workflows that classify requests and draft responses, supporting 400,000+ requests monthly with 93% resolution and 6.1x ROI.
• Built CaseLEA workflows for fraud and risk casework, gathering evidence across banking systems and preparing policy-checked cases for analyst approval.
• Scaled CaseLEA to 43,000+ cases processed, with average turnaround under 1 hour (down from 3+ days) and 98%+ closed within a 72-hour SLA.
• Built cross-system reconciliation workflows for a deployment handling 2M+ transactions daily, reducing effort by 80% with 2.7x ROI.
• Ran a ~10,000-run banking comparison across 150 scenarios: structured workflows reached 82% success versus 71% for autonomous agents.

Developed Smart Reports, the flagship product's first insights feature, combining AI workflows built with LangChain and statistical analysis of historical financial data to surface actionable findings.

Led frontend and DevOps development for a home care management and allocation platform, with contributions to backend services.

Developed the website and mobile app for the Quant-based Robo-advisory firm, offering AI and machine learning based portfolios. Led the switch to goal-based investing.

Developed the website for one of India's largest asset management companies, UTI Mutual Funds. Developed the banking website for HDFC, one of India's largest private banks.

Preference-Aware allocation and scheduling of teaching duties and through multi-constrained optimization