Abhishek Gupta

AI Engineer at droom.in (2025-10 – Present)

Designed a multi-agent AI system for rental, purchase, inspection orders, vehicle history, and valuation, using LLM-based intent classification to route user requests from an orchestrator agent to specialized sub-agents.
Improved response accuracy to 95% by building a RAG pipeline for company products, services, and user-guide queries, allowing AI agents to retrieve relevant data based on user queries and generate grounded responses.
Implemented Model Context Protocol (MCP) tools to expose backend APIs and databases, allowing agents to retrieve real-time data and execute relevant actions based on user queries.
Automated 20+ internal business workflows using n8n, integrating internal APIs, LLMs, scheduled triggers, Slack notifications, and email triggers for different business processes.
Delivered 95% MMYT (Make-Model-Year-Trim) matching accuracy by implementing a semantic matching service with vector search, LLM-based validation, and intelligent data normalization.

Software Engineer Intern - AI/ML at droom.in (2024-09 – 2025-09)

Reduced manual vehicle-inspection effort by 80% and achieved 90%+ accuracy by training a YOLO model on custom datasets for vehicle-view classification, part detection, and damage detection.
Implemented and maintained RESTful APIs to serve ML models and AI applications across multiple products, integrating them with internal services and optimizing response times.
Developed an OCR-based invoice processing solution that extracted customer and vehicle information, eliminating 100% of manual data entry and reducing processing time and errors.
Increased customer engagement by 45% by building WhatsApp and web chatbots for rental-order flows, integrating Meta WhatsApp Business APIs and backend services to automate customer interactions.
Achieved an R² score of 0.95 in repair cost prediction using a Linear Regression model trained on MMYT and part-level features, improving pricing consistency.

About