Aakash Maddheshiya

Architected and deployed an AI-powered duty reconciliation system to automatically validate US Customs duty sheets against supplier invoices, eliminating manual cross-verification and reducing processing time by over 60%. Implemented LLM-assisted semantic reasoning combined with rule-based arithmetic validation to ensure accurate verification of multi-component duty calculations (freight, insurance, customs duty, taxes). Designed and developed scalable Flask-based backend APIs to automate invoice ingestion, duty validation workflows, and financial accuracy checks across internal systems. Built data extraction and automation pipelines using Selenium to retrieve shipment and tracking data from courier vendor portals and government websites, improving operational visibility and reporting efficiency. Optimized validation workflows to minimize financial discrepancies and ensure compliance with customs documentation standards.

Developed intelligent document processing systems using Flask, OCR (PaddleOCR), and Computer Vision techniques to automate structured data extraction from invoices and enterprise documents. Integrated custom-trained YOLO (PyTorch) models into production-ready backend pipelines for image classification and object detection tasks. Designed and implemented embedding-based document retrieval systems using LangChain and LLM models, enabling semantic search and contextual Q&A over enterprise PDFs. Built modular and scalable REST APIs to support AI-driven document workflows and downstream integrations. Improved document processing accuracy and reduced manual intervention through AI-assisted validation logic.

Designed and deployed machine learning models for enterprise email classification and automated ticket routing using Azure ML Studio and GPT-based models. Built end-to-end AI automation workflows integrated with enterprise ticketing systems to streamline support operations. Implemented MLOps practices including CI/CD pipelines, model versioning, and deployment lifecycle management for production ML systems. Managed structured data storage and analytics using Azure SQL to support AI-driven reporting and monitoring. Contributed to the enhancement of enterprise applications through AI-driven automation and intelligent workflow optimization.