DHRUVIN R GANDHI

^About

I am Dhruvin Gandhi

Actively writing code and solving problems.

Driven software engineer and researcher who believes that innovation and perseverance are the keys to transforming ideas into impactful solutions.

I see every project as an opportunity to push boundaries and drive innovation. My journey in tech is fueled by curiosity and a relentless pursuit of growth. From crafting AI-driven systems to optimizing distributed networks, I’m dedicated to creating technology that not only works but makes a difference. I thrive in environments where learning is constant and collaboration is key. Right now, I’m diving deep into the realms of NLP and distributed systems, mentoring aspiring engineers, and always looking for the next big challenge to tackle.

Areas of Interests:

C++
Distributed Systems
Natural Language Processing
Data Science

Download Resume Send Email

Employment History

Cisco

Amherst, MA | Jan 2025 - May 2025

Developing an agentic RAG framework that enhances LLMs with user-specific graphs to personalize text generation.
Incorporating reasoning into the RAG pipeline, enabling more efficient data retrieval for in-context learning of LLMs.
Established a baseline metric of LLM personalization through RAG, using LLaMA-3.1-8B and distilled DeepSeek-R1

Advanced Networked Systems Research Lab

Amherst, MA | Sep 2024 - Jan 2025

Implemented consensus algorithms (Lazy and Chain Replication) to enhance fault tolerance in the XDN system.
Proposed and built an optimized interoperability pipeline for Java and C# integration using CoreCLR and JVM.
Reduced the cross language call latency in this pipeline by 40% enabling XDN to support any consensus algorithm

Interactive Brokers

Greenwich, CT | Jun 2024 - Aug 2024

Developed the logic for the formal models of distributed trading protocols using P language, C# and state machines.
Tested for consistency, fault tolerance, and latency specifications on these models for 1000 users.
Identified a critical vulnerability in real-time data access for multi-regional clients due to asynchronous network delays.
Proposed a working solution with Logical Lamport clocks, ensuring 100% correctness across distributed environments.
Automated Java method invocation in a P runtime by implementing a C++ bridge, boosting efficiency by 90%.

Undergraduate Research Assistant

VJTI Mumbai, India | Aug 2022 - May 2023

Proposed an AI-based novel information extraction workflow for Dark Web corpus.
Prioritized retrieval of critical data from unstructured text on illegal activities, contributing to efficient Dark Web analysis.
Eliminated the need for manual data labeling for named entity recognition of unseen classes in Dark Web Corpus
Leveraged information in the ConceptNet knowledge graph, achieving a 29.2% increase in F1 score.

Morgan Stanley

Mumbai, India | May 2022 - Jul 2022

Created a full-stack internal web tool with 10+ features using Spring Boot, Angular, TypeScript from scratch.
Addressed the challenge of unauthorized database updates through this tool which is now being utilized by 7 teams.
Built a reusable Angular UI for efficiently filtering amongst 10,000+ rows to perform visual CRUD operations.
Optimized Python script for syncing DEV Table with ∼28M rows, reducing execution time from 8hrs to 15 mins.
Collaborated with the finance team to understand requirements and include their feedback in development process.

Jobdae (Startup)

Remote, India | May 2021 - Jul 2021

Designed and implemented back-end web architecture using Python, Django, and MySQL
Created REST APIs defining functionality for profile management and job application tracking.
Developed an automated email notification system that promptly informs clients of job application submissions and status updates in real-time, boosting user engagement.

Projects

Context Aware Visual Question Answering

Identified key segments in images relative to textual queries by late fusion of textual embedding in a frozen Vision encoder. Fine-tuned a ViLT model for 3 epochs on Nvidia A100 GPU resulting in 12% increase in accuracy.

Code Paper

Personalizing LLM using Generative Retrieval

Generated hypothesis documents based on user input to improve the dense retrieval search in user profile history. Deployed a BERT model to extract relevant chunks, increasing LLM output personalization by 17%.

Code Paper

Fault-Tolerant Distributed Datastore

Implemented a multi-threaded Fault-tolerant replicated datastore algorithm using Zookeeper as the centralized service, ensuring consistency across replicated servers experiencing 1000 requests/sec from numerous clients.

Code Java, ZooKeeper, Cassandra

Farm Mart

Developed an eCommerce platform for farm-fresh produce featuring a chatbot on the homepage, product filtering, search, pagination, cart with discounts, Stripe API integration, and a blog with Buzzfeed-style quizzes using the News API.

Code NodeJS, JavaScript

DocuCare

Led a test-driven development of a MVC web application using agile software design principles such as, version control with Git, organizing development into sprints with fixed durations of 2 weeks and conflict resolution.

Code SpringBoot, Angular, Postgres

HisabKitab

HisabKitab is a simple GST billing and stock management software designed for local vendors and shopkeepers, offering easy access via mobile app or website to track transactions and print bills, bridging the tech gap between small and large-scale vendors.

🥈 2nd Prize - Hackathon

Code Flutter, Firebase, NodeJS

More Projects on Github

Education

Master of Science in Computer Science

University of Massachusetts Amherst

Aug 2023 - May 2025 (Expected)

GPA: 3.95/4.0
Relevant Courses: Machine Learning, Distributed Computing and Systems, NLP, Information Retrieval, Data Science Fundamentals, Software Engineering

Bachelor of Technology in Computer Engineering

Veermata Jijabai Technological Institute (VJTI), Mumbai

Aug 2019 - May 2023

GPA: 9.13/10
Relevant Courses: Data Structures, Algorithms, Operating Systems, Database Systems
Final Project: "AI-Based Information Extraction System for Dark Web Data"

^Testimonials

Kind Words
from Colleagues

“Curious Researcher”

Dhruvin worked as a 2024 summer intern at Interactive Brokers. He was an enthusiastic communicator and a curious researcher. During the summer, he required little supervision and can perform tasks independently. At the end of his internship, he successfully presented his work on formal modeling using P-Language.

Yi Suo

Interactive Brokers

“Team Player”

Dhruvin holds a passion for software development. Though I didn't manage Dhruvin directly, I still know him because the best team of software engineers worked for Jobdae that indicates that he is a team player. The skills and passion he has reflected in his work as timely delivery of work and problem-solving skills.

K. Pooja

Jobdae

“Problem solver”

Dhruvin was a great professional to work with. We worked together at Jobdae and he helped me a lot throughout the internship. He always found great ways to quickly solve problems and his deliveries were always very well executed, on time and flawless. I’d definitely like to work with him again.

Praveen Yadav