Hi, I'm Kunj Joshi

Computer Science Student

I am a Master's Student in Computer Science at
Khoury College of Computer Sciences
Northeastern University, Boston.

Introduction

I am a Graduate Student of Computer Science at Northeastern University, Boston. I am highly interested in researching security measures in Generative AI and LLMs. With my work experience in both Generative AI and Fullstack Development, I strive forward to make an Augmentative AI: An AI that focuses on helping humanity rather than replacing it. I am proficient in various fields of Software Development including Backend Engineering and Cloud Deployment. My programming experience spans over a wide array of Frameworks and Languages, most notably Python (Django, Flask and FastAPI), JavaScript (ReactJS and React Native) and Java (JavaSwing, Hibernate). I love writing blogs and poetries and hiking during my free time.

About Profile Image Headshot for Kunj Joshi

Stats

Academic Performance

Master's GPA: 3.93/4.00

Bachelor's GPA: 9.67/10.00

Professional Experience

Total Years: 2+ Years

Full-stack Development, AI/ML Engineering, and Research

Projects Completed

Total Projects: 6

Research Publications

Total Publications: 3

Including ArXiv Preprints

Skills

Programming Languages

  • Python
  • JavaScript (ES6+)
  • Java
  • TypeScript
  • C++
  • SQL
  • Golang

Web Frameworks

  • Django
  • Flask
  • FastAPI
  • ReactJS
  • React Native
  • Node.js
  • Express.js

Database Technologies

  • PostgreSQL
  • MySQL
  • MongoDB
  • Redis
  • Firebase
  • Hibernate

Tools and Platforms

  • Git & GitHub
  • Docker
  • AWS (EC2, S3, Lambda)
  • Google Cloud Platform
  • Linux/Unix
  • VS Code
  • Postman

AI/ML Technologies

  • TensorFlow
  • PyTorch
  • Scikit-learn
  • OpenAI API
  • Hugging Face
  • LangChain
  • Natural Language Processing

Education

Pandit Deedayal Energy University

Degree: Bachelor of Technology in Computer Science

Duration: 2019 - 2023

GPA: 9.67/10.00

Key Achievements:

  • Subjects studied: Operating Systems, Computer Networks, Algorithms, Data Structures Cybersecurity, Information Security, Machine Learning, Data Mining
  • Worked as a Research Assistant in Cybersecurity under Professor Kaushal Shah and Professor Chintan Bhatt and published two publications in a renowned journal and a conference.

Northeastern University

Degree: Master of Science in Computer Science

Duration: 2024 - Present

GPA: 3.93/4.00

Focus Areas:

  • Subjects studied: Algorithms, Programming Design Paradigm, Cybersecurity Principles, Machine Learning, Advanced Machine Learning, Web Development
  • Working as a Research Assistant under Prof Udi Hoitash from DAmore McKim School of Business in Generative AI.
  • Conducted Research under Prof David Smith in Memorization of LLMs and proposed a novel mitigation technique that outperformed current benchmarks

Work Experience

Rewardwise - Fullstack Developer

Duration: Mar 2023 - Nov 2023

Responsibilities:

  • Led backend development for Rewardwise app using Python web framework Django and ReactJS; built modules for 4 key features: Ticket Booking, Sales Analysis, Rewards/Offers and Memberships.
  • Optimized user data extraction from an SQL-based database by implementing efficient query techniques and leveraging advanced data attributes such as time, location, and numerical metrics, enabling seamless processing for over 100,000 users.
  • Implemented controlled CI/CD deployment with GitHub Actions on AWS EC2, using YAML to deploy with less than 0.3 ms latency

Nirnaya Systems - AI Engineer

Duration: Nov 2023 - June 2024

Responsibilities:

  • Improved transformer-based NMT and QA models, increasing accuracy by 12.65 BLEU points using LoRA and PEFT
  • Boosted sentence similarity vector scores by over 78% on live datasets for efficient Retrieval Augmented Generation using Python Pandas Dataframe, CSV file formats and vectorization using VectorDB
  • Spearheaded Document Layout Analysis and Image Noise Reduction Solutions, improving existing models by 12.34% in Border Detection Accuracy and Image Smoothening and deployed using FastAPI Backend and ReactJS Frontend, with quantized model deployment.

Northeastern University - Research Assistant

Duration: Feb 2025 - Present

Responsibilities:

  • Designed 3 RAG pipelines using Mistral and GPT models with vector indexing for change detection across 50,000+ documents spanning over 18 years, implementing hybrid search and semantic embeddings, improving cosine similarity accuracy by 46%.
  • Spearheaded the enhancement of Text Extraction and Vectorization methods, achieving a 65% improvement by introducing archi- tectural changes in Language Models with seamless deployment using Kubernetes and TensorRT
  • Optimized deployment using 8-bit quantization and WeaviateDB Vector Store, reducing GPU memory usage by 24.67% while maintaining 100 queries/sec throughput with MLOps monitoring.

Traderware - Generative AI Intern

Duration: June 2025 - Aug 2025

Responsibilities:

  • rchitected TraderGPT multi-agent system with 5 specialized LLMs and 15+ financial tools through MCP pipeline, imple- menting Orchestrator for routing, dual SQL LLMs for query generation, Financial Analyzer for forecasting, and Aggregator for synthesis—validated by 3 senior analysts for stock predictions.
  • caled to 18+ distributed LLM system with 50+ tools using LangGraph, optimizing inference throughput for 5-minute end-to- end latency on complex multi-step queries requiring PostGreSQL and GraphQL analysis and time-series forecasting.
  • Created real-time news clustering with DBSCAN and MiniLM embeddings, processing 10,000+ articles into 200+ clusters based on 44 features, dispatched with Docker Containers, Github Actions and YAML CI/CD Pipeline.

RoonCyber - SDE Intern

Duration: Sept 2025 - Dec 2025

Responsibilities:

  • Fine-tuned LLM on 20,000+ Python code snippets using LoRA and PyTorch with distributed training, achieving 94% accuracy in detecting 750+ vulnerabilities across 15+ categories with LLM Guardrails to prevent malicious code generation
  • Built Anomaly Detection System (ADS) for AWS cloud logs, processing 10,000+ daily security events and classifying into 5 priority tiers using LangChain, identifying 3 critical vulnerabilities missed by CodeRabbit and GitHub Copilot.
  • Devised automated retraining pipeline for Microsoft Phi-4 using Apache Airflow, on AWS CloudWatch logs every 50 hours, improving ADS detection accuracy by 12% and reducing false positives

DASH Lab - Graduate Researcher

Duration: January 2026 - Present

Responsibilities:

Carousel of Photos

San Francisco Biking Selfie
Trekking at Artists Bluff New Hampshire
Learning Basketball in Early 2025
Times Square At New York City on New Years Eve
Favorite Photo Location on Boston Seaport

Contact Form

Submit

Your information won't be shared. Never submit passwords.