AI Engineer at Cadence Premier Logistics

Architected a dual-facing multi-agent system with LangGraph to automate driver support and safety escalation workflows, integrating Model Context Protocol (MCP) to connect LLMs with real-time internal databases and logistics tools. Built a RAG pipeline using LangChain and pgvector to index internal safety manuals and compliance documents, improving first-response accuracy and reducing escalations to human dispatchers. Engineered a serverless data integration pipeline using AWS Lambda and EventBridge to sync 1,000+ driver records daily between internal systems and Monday.com CRM, reducing execution time by 70% and eliminating 100 hours/week of manual data entry. Automated invoice processing workflows using n8n, JavaScript, and AWS Textract, extracting and validating 200+ invoices weekly while reducing manual reconciliation time by 60%. Deployed and monitored production LLM workflows on AWS Lambda and SageMaker with Docker, using LangSmith for tracing and evaluation while sustaining 99%

Graduate Teaching Assistant - Software Modeling Development at Illinois Institute of Technology

Delivered hands-on tutorials and lab sessions on software system design, modeling, and development practices to 70+ graduate students. Covered UML modeling, design patterns, and software architecture to strengthen comprehension and project outcomes. Developed a review recommendation engine demo using the Yelp Fusion API, OpenAI embeddings, and Elasticsearch to demonstrate embedding-based semantic search and retrieval ranking. Containerized the application using Docker to teach deployment patterns and container-based workflow practices.

Software Engineer Intern at CDK Global

Developed a scalable billing dashboard on the Fortellis platform using React for 500+ software vendors to monitor API consumption and subscription billing data totaling $2M+ in transaction volume. Architected asynchronous RESTful backend services in Node.js to improve response times by 30% and handle 40,000+ billing records stored in DynamoDB. Integrated Okta SSO authentication with role-based access controls across enterprise applications. Automated build and deployment pipelines with GitHub Actions, executed unit testing, and collaborated in Agile sprints using Jira to improve release reliability and accelerate feature delivery.

Software Engineer - AI/ML at Bhaskaracharya Institute for Space Applications and Geo-Informatics (BISAG)

Fine-tuned a pre-trained Vision Transformer (ViT) with PyTorch on satellite imagery datasets to improve generalization and inference performance for land use and geospatial analysis. Built automated data preprocessing and feature engineering pipelines in Python to optimize geospatial and agricultural datasets for LightGBM training, reducing data preparation time by 30% and improving reliability. Designed a PostgreSQL database schema using PostGIS extensions to store and query geospatial datasets and classification outputs. Deployed trained models as REST APIs using FastAPI for real-time land classification and yield predictions, reducing average query latency to under 200ms. Created a Streamlit dashboard integrated with Elasticsearch-indexed survey reports to visualize model outputs alongside historical land-use records and reduce manual document lookup time by 50%.