Kishor Sarkar

Leading frontend and backend engineering across AI model training and mortgage platforms, delivering annotation workflows, data quality systems, and micro-frontend architectures.

• Built annotation workflow interfaces and BullMQ-backed validation pipelines that reduced invalid data entering AI training pipelines by ~30%.
• Engineered data quality systems using Survey JS, React, Node.js, and PostgreSQL — enabling reliable benchmarking at scale across structured annotation tasks.
• Architected a micro-frontend system for admin, broker, and customer portals — reducing deployment time by 40% and enabling independent team releases.
• Led WebSocket integration and Lighthouse-driven performance audits, improving real-time loan status visibility and reducing page load times across the mortgage portal.

Built the operational admin portal for Dealyards from scratch, centralizing business workflows and replacing fragmented manual reporting with live real-time data.

• Developed the full admin portal from the ground up using Angular, Node.js, and Socket.io.
• Replaced manual, fragmented reporting with a real-time data dashboard, streamlining core business operations.
• Implemented live Socket.io event streams for instant operational visibility across the platform.

Developed a centralized fraud detection platform for IACC, consolidating counterfeit and duplicate listing investigations across multiple e-commerce marketplaces.

• Built the fraud detection platform using Angular and Node.js with cron-based scrapers targeting multiple e-commerce marketplaces.
• Optimized frontend architecture for performance and scalability — enabling the platform to handle growing data volumes without workflow degradation.