Software Development Engineer at Eminent Services Corporation

Frederick, MD

Project: Investigational Drug Management System Modernization involved transforming a critical VB6 application that was becoming increasingly difficult to maintain and scale. The project required analyzing the existing system, designing a modern architecture using React & Node, and ensuring zero data loss during migration. The modernization resulted in 35% improved scalability through better resource utilization and 40% reduction in maintenance efforts due to modern tooling and clear separation of concerns.

As a Software Development Engineer I:

• Converted legacy VB6 application into MERN stack with modular React components, improving scalability by 35% and reducing maintenance efforts by 40%.

• Built cloud-native deployment pipeline using Azure Pipelines and App Service, reducing deployment time by 30%.

• Designed data migration plan with schema evolution and event logging, reducing manual cleanup by 75%.

ML Research Assistant at University at Buffalo

Buffalo, NY

Project: Energy Efficiency in Large Language Models focused on optimizing energy efficiency in LLMs, specifically GPT-2, to address the growing environmental impact of AI model deployments. The research involved implementing and evaluating various optimization techniques including quantization, knowledge distillation, and low-rank factorization. Tools and libraries including PyTorch, Transformers, Accelerate, and Auto-GPTQ were utilized to measure and track CO2 emissions, power consumption, and energy metrics during model training.

As a ML Research Assistant I:

• Reduced GPT-2 energy consumption by 19.8% through quantization techniques while maintaining model accuracy.

• Explored combined distillation and quantization methods, achieving 45.2% lower emissions in language model training.

• Published findings in departmental research paper and presented optimization techniques at graduate research symposium.

Software Development Engineer at Swym Corporation

Bangalore, Karnataka, India

Project: Wishlist Plus & Back in Stock Alerts are powerful e-commerce apps that enhance the shopping experience and drive sales. Wishlist Plus enables personalized wishlists with cross-device access, customization, and social sharing, providing valuable insights. Back in Stock Alerts allows customers to receive notifications for out-of-stock or pre-order items, maximizing sales potential with advanced controls and SMS support. Both apps facilitate social re-engagement, targeted emails, and list growth, offering businesses actionable insights and analytics for optimal performance.

As a Software Development Engineer I:

• Engineered and customized 30+ bespoke wishlist solutions for e-commerce stores by leveraging REST APIs, thereby optimizing user journeys and aligning with diverse business requirements.

• Innovated and implemented an internal knowledge base tool, empowering Swym's development and support teams with a comprehensive resource for quick reference, fostering streamlined collaboration and problem-solving.

• Resolved 8+ caching issues by optimizing in-memory caching on the server-side, resulting in a 12% improvement in response time. This enhancement facilitated faster real-time updates, contributing to an enhanced user experience.

• Revamped external developer documentation by incorporating code recipes and demos, enriching the content for efficient utilization of REST APIs. This initiative led to a 10% increase in traffic to the developer docs.

Research & Development Intern at Samsung PRISM

Bangalore, Karnataka, India

Project: Gateway for supporting IMS signalling. The primary objective was to address the interoperability challenge between two distinct realms of communication technology: VOIP controlled by Over-The-Top (OTT) players and Voice over LTE (VoLTE) managed by telecom operators. The focus was on creating a gateway that could seamlessly handle websocket SIP requests, transform them into UDP/TCP or TLS, and vice versa, facilitating communication between these two worlds.

As an Intern I:

• Conceptualized an IMS signaling gateway between PC and mobile devices to provide simple end-to-end video communication.

• Formulated and constructed a pipeline for the gateway to convert TCP/IP packets into SIP Packets and vice-versa to make them both interoperate.

• Incorporated the use of WebRTC stack present in modern web browsers for enabling audio/video real-time communication.

Technology Intern at Head Digital Works Pvt Ltd

Hyderabad, Telangana, India

• Developed an application that helps to automate the data flow between several Cassandra nodes and application endpoints with minimal data loss and higher data transfer rate.

• Structured and Implemented GraphQL queries on top of existing REST API to avoid under-fetching and over-fetching of data.

• Coordinated with server-side by utilizing Cassandra on AWS, coupled with Node.js, and Express.js for client-side development