Deepak Raj

Senior Software Engineer

• Document Intelligence — Increased document processing efficiency by implementing a YOLOv5 object detection model within the OCR pipeline, achieving 97% accuracy in document identification.
• Claim Automation — Automated the insurance claim process and reduced human involvement, cutting processing time from 21 days to 3–4 days.
• Knowledge Graph & LLMs — Developed an advanced solution leveraging Abstract Syntax Trees (AST), Neo4j, LangChain, and Large Language Models (LLMs) to structure, store, and query insurance data, enabling end users to obtain accurate answers to insurance-related queries.
• Reverse Engineering — Automated insurance product illustration generation through reverse engineering, eliminating legacy code and boosting processing speed.
• Equity Research Automation — Automated the equity research process by extracting data from documents, mapping relationships, and generating industry-standard reports — reducing turnaround time from 3 days to 5 hours.
• Legacy Modernisation — Migrated legacy Progress code to modern Java and Python; modernised a 20-year legacy application using NestJS and Next.js, deploying a high-performance B2B API.
• Document Extraction — Built a document extraction system by combining two complementary algorithms, leading to more accurate and reliable results.
• Computer Vision for Industry — Worked on computer vision use cases for industrial sites to strengthen security protocols and improve workplace safety.
• Mentorship — Mentored junior developers on modern software design and architecture practices, helping build a team culture focused on learning and growth.

Associate Data Scientist

• ML APIs & Pipelines — Built robust APIs and data pipelines using FastAPI and Flask to handle configuration, deployment, and benchmarking of ML/AI models.
• Dashboards & Reporting — Developed and maintained interactive dashboards and reporting tools that improved tracking of targeting models and enhanced management of key business metrics, leading to more informed decision-making.
• Client Reporting — Delivered visually impactful web dashboards and client reports using Python (Dash/H2O), Tableau, and PowerPoint, providing insight into product output, performance, and user behaviour.

AI Developer

• Computer Vision & NLP — Designed, developed, and implemented computer vision and NLP solutions using deep learning frameworks for mobile-first deployment.
• Mobile AI Integration — Worked closely with the Android development team to deploy AI applications into production mobile environments.
• Stakeholder Coordination — Coordinated client appointments and supported executive meetings to ensure smooth project delivery.

IMS ProSchool

University of Delhi