Ramandeep Sharma

About Ramandeep Sharma

I build intelligent, agent-based automation in Microsoft Copilot Studio—integrating Dynamics 365, Power Platform, and backend services through the Model Context Protocol (MCP). My current focus is on designing the Derivation Agent and OCR Agent that will modernize and ultimately replace the existing Invoice Capture application. This includes building scalable MCP servers, intelligent extraction and derivation workflows, and LLM-driven decisioning patterns for AP automation.Previously, I was a Principal Software Engineer at Oracle, where I led the design and development of resilient cloud-native microservices on Java, Spring Boot, and OCI. With 13+ years of engineering experience, I specialize in distributed systems, backend architecture, reliability engineering, and building platform capabilities that scale globally.I’m passionate about solving complex problems with clean system design, automation, and continuous innovation—while driving clarity, collaboration, and technical excellence across teams.

Experience

1. Senior Engineer

   Microsoft

   Aug 2024 — Present

   Architecting and delivering a next-generation AI-powered Accounts Payable automation platform processing invoices per day within Microsoft Dynamics 365 Finance, using Microsoft Copilot Studio and Model Context Protocol (MCP). Leading development of Derivation Agent and OCR Agent within the D365 Finance Invoice Capture module, replacing a legacy system with LLM-driven decisioning, high-accuracy OCR pipelines, and fuzzy matching models. Designing and scaling multi-tenant MCP servers to support high-volume document ingestion, extraction, validation, and orchestration workflows at enterprise scale. Built an end-to-end derivation framework enabling AI-based field inference, rule evaluation, validation, and consistency checks across financial documents. Led development of agent lifecycle infrastructure including evaluation and feedback loops using Braintrust-based evaluation pipelines, improving LLM output quality and system accuracy. Drove performance and reliability improvements across financial workflows — optimizing distribution splits, tax calculation, and large-invoice processing pipelines. Led cross-team architectural alignment across Accounts Payable, Source Document, and MCP platform teams, influencing long-term system design and integration strategy. Defined and drove the technical roadmap for transitioning Invoice Capture to an agent-based architecture, aligning multiple teams and long-term platform evolution. Mentored engineers and conducted architecture reviews, improving debugging practices, system observability, and distributed system reliability.

Education

• SSGPURC, Punjab University

  BE, Computer Science

  2008 — 2012

• Birla Institute of Technology and Science, Pilani

  Master of Technology - MTech, Artificial intelligence and machine learning

• Punjab University

  Bachelor of Technology

Skills

• Java Enterprise Edition
• Jquery
• Web Applications
• Tomcat
• Maven
• Angularjs
• Mvc Architecture
• Databases
• Mvc
• Sql
• Agile Application Development
• Ajax
• Mysql
• J2ee Web Services
• Software Development
• Json
• Css
• Javascript
• Java
• Cassandra