◦ Optimized Artifact Search: Improved artifact retrieval by enhancing OpenSearch indexing pipelines, reprocessing

existing data via migration scripts, and implementing relevance-based filtering in the UI, reducing search response

time by 35%.

◦ Developed Dynamic Form-Based Workflow Engine: Implemented a template builder with configurable input

components (text, uploads, selections, signatures) using React, Node.js, AWS Lambda, and DynamoDB, enabling

admins to create reusable task templates and reducing manual workflow setup time by 40%.

◦ Enhanced AI Knowledge Assistant TwinGPT (RAG): Built document ingestion and semantic retrieval APIs

using Pinecone and Gemini/OpenAI, enabling users to query manuals conversationally and improving documentation

accessibility by 55%.

◦ Built Data Ingestion Pipeline: developed an Excel-based ingestion service to process assets and parts data into

Amazon Neptune (graph DB) using Gremlin queries, orchestrated via Durable Functions and AWS Lambda handlers,

enabling automated data creation and reducing manual data entry efforts by 70%.

◦ Developed a scalable report generation system using Handlebars driven HTML partials, integrated with a form

capture service and DynamoDB, and implemented a PDF generation engine to resolve dynamic data reducing report

turnaround time from hours to minutes and improving operational efficiency by 65%.

◦ Engineered DocGPT – AI Document Analysis System: developed a platform to process bulk inspection,

device-report PDFs, extract key defects using Gemini API, and present structured findings with PDF highlight

navigation, reducing manual review time by 80%.

◦ Built PDF Annotation Extraction Service: Developed an AWS Lambda-based solution using PyMuPDF to

detect circular annotations and extract associated text with precise bounding boxes, enabling interactive PDF overlays

in React and improving part identification efficiency by 40%.

◦ Resolved DynamoDB storage bottleneck: using zlib compression, shrinking large task payloads by 96% and

enabling seamless handling of large form submissions without schema or infrastructure changes.

◦ Engineered a GenAI summarization system: with a centralized prompt library mapped to task types,

generating AI-powered summaries from user task data via OpenAI/Gemini and persisting results in-app — enabling

fully configurable, zero-duplication prompt management across workflows.

◦ Tech Stack: TypeScript, JavaScript, Python, React, Node.js, DynamoDB, Amazon Neptune, Amazon

EC2, Amazon S3, AWS Lambda, Amazon CloudWatch, Serverless Framework, GitHub, Material UI

(MUI)