Designed and deployed a Retrieval-Augmented Generation (RAG) system for legal documents using Typescript,

Node.js, OpenAI APIs, and FAISS

• Built prompt templates, chaining logic, and evaluation pipelines to improve LLM response accuracy for legal

summarization workflows.

Working as a Software Engineer/Research Assistant at the MICV3D lab under Dr. Christopher Lauren at Purdue University, Indianapolis.

Key Responsibilites:

• Conducting research in Computer Vision, leveraging CNNs, Vision Transformers, and image representation learning.

• Designed and developed a mobile app in flutter to capture pictures under different LED settings and frequencies to see redacted documents.

• Integrated Audrino to control the frequencies from the mobile app.

• Augmented Reality functionality to detect objects and take measurements in real time.

• Fingerprint Identification functionality using YOLOv8 and Roboflow to detect fingerprints in the images.

• Used ONNX to convert the pytorch model to tensorflow-mobile for deploying into the flutter app.

• Achieved a precision score of 99.5% and 84.27% recall.

Designed and developed microservices supporting business verticals like Insurance, Medicine Delivery, and Platform Services with high availability and fault tolerance.

Key Contributions:

• Database optimization project across the organization, leading to the strategic downgrade of the AWS SQL server instance from 24x to 8x. Achieved a substantial cost reduction of approximately 100k USD anually.

• Integrated insurer APIs and webhooks, implemented data encryption, and deployed serverless functions on AWS Lambda to ensure secure and scalable system operations.

• Designed and developed RESTful APIs for an order management system for diagnostic centers and insurers,enabling seamless order processing and customer report management, handling over 10,000 orders per day.

• Redesigned the zip code–based order booking system at Medibuddy using a geolocation approach with latitude/longitude, MySQL, and the Google Distance Matrix API, reducing user fulfillment time by 30%.

• Led monolith to microservices migration, improving system latency by 40% and ensuring CI/CD integration using tools like Docker and Jenkins.

• Streamlined production monitoring by integrating New Relic, Elastic Search, and AWS Kinesis, decreasing mean time to resolution (MTTR) for critical issues by 60% and improving overall system reliability.

• Diagnosed and fixed production issues involving slow database queries, read replica lag affecting order status updates, and cache invalidation, improving system reliability and reducing downtime.

Worked as a Software Engineer on Intel's Enterprise Data Platform team. I contributed to the development of a Conversational AI platform enabling various Intel product groups to create chatbots with minimal effort and no development expertise required.

Key responsibilities:

• Design and developed RESTful APIs, deployed on Cloud Foundry, supporting scalable and reliable backend services..

• Improved chatbot accuracy by 30% by optimizing Rasa NLU pipelines and implementing a custom dictionary for Intel-specific acronyms.

• Built scalable and reusable UI components in Angular, enhancing rendering performance and accelerating feature development.

• Migrated and automated CI/CD pipelines from TeamCity to GitLab, cutting licensing costs by 40%.

• Developed an automated code scanning system using AppSec APIs to identify vulnerabilities in GitLab deployment pipelines to production.

• Wrote unit tests using Mocha and Chai, increasing test coverage by 90% and ensuring higher code reliability.

• Automated Intel’s access control management, reducing the SLA for project creation from hours to 30 minutes.

• Resolved critical production issues using ServiceNow ensuring minimal downtime and continuous business operations.