Arpit Vaghela

AI/ML Engineer | Full-stack Developer
Toronto, CA.

About

Highly analytical and detail-oriented AI/ML Engineer with a Master's degree in Computer Engineering, specializing in Large Language Models, full-stack development, and cloud deployment. Proven ability to deliver high-impact technical solutions, from developing custom watermarking pipelines for LLMs to building scalable web applications with a 99.9% uptime, driving significant improvements in product stability and user engagement.

Work

University of Guelph
|

Research Assistant - Watermarking for Large Language Models (LLMs)

Guelph, Ontario, Canada

Summary

Conducted advanced research on text watermarking for LLMs, developing and evaluating novel hybrid strategies to enhance AI content traceability and plagiarism detection.

Highlights

Pioneered research on text watermarking methods for LLMs across 3 core NLP tasks (summarization, QA, generation), focusing on attribution, plagiarism detection, and AI content traceability.

Engineered a custom PyTorch/Hugging Face watermarking pipeline, fine-tuning 5+ transformer models with minimal loss in fluency (less than 5% degradation in ROUGE and perplexity).

Evaluated trade-offs between watermark detectability and task accuracy using BLEU, ROUGE, and perplexity metrics, developing and comparing both deterministic and probabilistic watermarking strategies.

Co-authored a research paper and developed a reproducible testing framework, advancing academic work on AI safety and content authentication.

Engineering Ambition
|

Full stack Developer and UI / UX Designer

Guelph, Ontario, Canada

Summary

Led full-stack web application development and UI/UX design, significantly boosting product stability and user satisfaction through robust architecture and efficient AWS deployments.

Highlights

Developed and maintained full-stack web applications using Next.js, Node.js, and MongoDB, achieving a 25% boost in product stability and 2x faster page load times.

Designed 15+ responsive UI components with Figma and Adobe XD, enhancing user satisfaction by 20% across desktop, tablet, and mobile platforms.

Managed AWS deployments and scaling, reducing server response time by 30% and ensuring 99.9% uptime.

Coordinated Agile sprints via ClickUp, achieving a 90% task completion rate and a 30% drop in bug reports through active code reviews and GitHub workflows.

Credence Technologies
|

Web Developer Intern

Surat, Gujarat, India

Summary

Designed and developed responsive web applications, enhancing user engagement and optimizing backend performance and database efficiency.

Highlights

Designed and developed responsive web applications with MERN stack, boosting user engagement by 25% through optimized design and functionality.

Collaborated with backend developers using Node.js and Express to connect APIs and server-side functions, reducing page load times by 15%.

Optimized SQL and MongoDB database schemas, ensuring efficient data retrieval and storage, which improved overall application performance and reliability.

Google Developer Student Club
|

Core Member

Surat, Gujarat, India

Summary

Orchestrated and led technical workshops and events, supported Google technology presentations, and mentored junior members, fostering skill development and a culture of continuous learning.

Highlights

Organized and led workshops, hackathons, and coding competitions over six months, significantly contributing to student learning and skill development.

Supported the development and delivery of interactive presentations on Google technologies (Android, web, cloud, ML), effectively engaging attendees and improving their understanding.

Mentored and guided junior members for five months, promoting their growth and active participation in club operations, fostering a culture of continuous learning and collaboration.

Education

University Of Guelph
Guelph, Ontario, Canada

Master of Engineering

Computer

Sarvajanik College of Engineering and Technology
Surat, Gujarat, India

Bachelor of Engineering

Computer

Awards

Larsen and Toubro Scholarship

Awarded By

Larsen and Toubro

Awarded for maintaining a high GPA throughout each semester of undergraduate studies, demonstrating academic excellence and commitment to education.

Skills

Programming Languages

Python, SQL, JavaScript.

Machine Learning & Deep Learning

PyTorch, TensorFlow, Keras, Scikit-learn, Transformers, Hugging Face Spaces, Gradio, LLMs, NLP, Retrieval-Augmented Generation (RAG).

Data Analysis & Visualization

Pandas, NumPy, Matplotlib, Seaborn, Jupyter Notebook.

Cloud & Deployment

AWS S3, AWS EC2, AWS SageMaker, Docker.

Version Control

GitHub, Git.

Web Technologies

Next.js, Node.js, MongoDB, MERN Stack, Express, Figma, Adobe XD.

Project Management

Agile, ClickUp.

Projects

Breemary

Summary

Fine-tuned Meta's BART-Large-CNN model for multi-source text summarization, optimized for performance and deployed via Gradio.

WhatToCookToday

Summary

Created a smart recipe recommendation system using RAG and semantic search, deployed as an interactive cooking assistant UI.

Foodvision

Summary

Developed a food classification model using PyTorch and Vision Transformer (ViT) architecture, deployed via Gradio.

Dogvision

Summary

Developed a high-speed, lightweight dog breed classification model using PyTorch, deployed via Gradio.