Parth Vashisht

About Me

I am a Machine Learning Engineer with a passion for developing and implementing cutting-edge AI solutions. With a strong background in computer science and extensive research experience, I specialize in machine learning, natural language processing, and multimodal systems.

Skills
Natural Language Processing Research100%
Python 95%
Machine Learning 85%
Pytorch 90%

Resume

My professional journey and academic background
Download my resume

Education

Masters of Science in Computer Science

Sep 2022 - May 2024

University of Massachusetts Amherst, MA

Courses: Machine Learning, Natural Language Processing, Distributed Systems, 3D Computer Vision

Bachelor of Science in Computer Science

Aug 2016 - Jul 2020

Chitkara University, India

Courses: Data Structures and Algorithms, Web Development, Probability and Statistics, Artificial Intelligence

Professional Experience

Machine Learning Intern

Jun 2024 – Present

Vectara, Palo Alto, California

  • Optimizing Vectara’s Boomerang model for improved multilingual text retrieval.
  • Enhancing re-ranking performance across diverse retrieval datasets.

Graduate Student Researcher

Jan 2024 – June 2024

Allen Institute for Artificial Intelligence, Amherst, MA

  • Developed a reinforcement learning framework for enhancing moral reasoning tasks in LLMs.
  • Finetuned open-source models to align with human moral reasoning.

Graduate Student Researcher

May 2023 – Present

BioNLP lab, University of Massachusetts Amherst, MA

  • Led research on synthetic data generation for clinical note summarization using GPT-4 and GPT-3.5.
  • Improved Visual Question Answering task accuracy by 23% using multimodal RAG pipeline.

Software Development Engineer

Oct 2019 – Feb 2021

FreightDeck Technologies, Bengaluru, India

  • Enhanced network performance using Socket.io with Node.js and Express.js backend.
  • Implemented Redis caching and AWS S3 integration to improve server response time by 40%.

Research & Projects

My latest Research Work and Projects

Research Publications (EMNLP, NAACL, Springer)

SYNFAC-EDIT: Synthetic Imitation Edit Feedback for Factual Alignment in Clinical Summarization*

In Review EMNLP - 2024

Performance of Multimodal GPT-4V on USMLE with Image: Potential for Imaging Diagnostic Support with Explanations

In Review Springer Nature - 2024

UMass-BioNLP at MEDIQA-M3G 2024: DermPrompt - A Systematic Exploration of Prompt Engineering with GPT-4V for Dermatological Diagnosis

Accepted NAACL Clinical NLP Workshop - 2024

Projects

Abstract Artwork Image Captioning

Deployed BLIP for the image captioning task, applying Supervised Fine-Tuning and leveraging PEFT techniques using LoRA to maximize image-text correlation and improve caption quality by 13% points using GPT4-Evaluation.

Multimodal Misogyny Classification

Modified the architecture of CLIP model by adding multiple fusion points and finetuning the model using custom loss to achieve state of the art multi-label classification accuracy.

Contact

Get in touch with me

Location:

Palo Alto, California