Parv Bhargava

MS in Data Science

August 2023 - Present

Relevant Coursework

• Deep Learning
• Machine Learning
• Data Mining

B.Tech. Computer Science and Engineering

July 2019 - May 2023

Relevant Coursework

• Database Management Systems
• Data Structures and Algorithms
• Computer Vision

Temporai

May 2024 - Present

Data Scientist

• Led the end-to-end development of a multi-agent conversational system using AutoGen and LangGraph, seamlessly integrating a Retrieval-Augmented Generation (RAG) pipeline that reduced bot hallucinations by 95% and improved response time by 30% based on user feedback. Proactively led brainstorming sessions, fostering a culture of innovation and knowledge sharing.
• Engineered state-of-the-art intent classification models for efficient data retrieval from complex systems using natural language queries, significantly enhancing bot efficiency across 500+ pre-launch test cases.
• Achieved a 98% system success rate by orchestrating a multi-LLM inference strategy utilizing GPT-4, LLaMA, and Claude, while reducing false positives by 40% through strategic process improvements.
• Developed a domain-agnostic time series forecasting tool integrating statistical and deep learning models, achieving 93% accuracy. Implemented a stacking ensemble and optimized data preprocessing, reducing turnaround time by 50%.

The George Washington University

March 2024 - Present

Student Research Specialist III

• Led the development of a Hybrid RAG architecture using Neo4j knowledge graphs for context retrieval, enhancing retrieval efficiency and observability. Worked closely with stakeholders to gather requirements and ensure effective information access from documents.
• Designed and deployed a chatbot for policymakers to conduct research with high accuracy, reducing LLM hallucinations and improving information reliability. Enabled stakeholders to efficiently access sourced information from documents.

The George Washington University

January 2024 - Present

Graduate Assistant

• Managed AWS infrastructure for university courses, deploying Lambda functions institution-wide, resulting in a 35% reduction in operational costs. Processed and prepared large datasets for image classification tasks.
• Configured and optimized CUDA and cuDNN environments for TensorFlow and PyTorch on AWS EC2, enhancing the performance of deep learning models. Took initiative to document best practices, enabling future teams to set up efficiently.

MunshiG

April 2023 - June 2023

AI Intern

• Integrated GPT-4 model with advanced prompt engineering and function calling to build a multi-functional AI agent deployed across multiple social media platforms, enhancing user engagement and automation.
• Set up a NoSQL database using MongoDB for efficient and secure data handling, improving data retrieval speed by 40%.

Appronic

August 2022 - March 2023

Data Scientist

• Developed a scalable data processing pipeline using Apache Spark and optimized SQL queries, reducing processing time by 30% and enabling real-time analytics. Worked closely with data engineers and analysts, ensuring data integration.
• Built time series models that improved forecast accuracy by 15%, delivering reliable data-driven insights.
• Implemented a decision tree-based machine learning model for customer churn prediction, increasing prediction accuracy by 20% and enabling targeted retention strategies. Worked in close collaboration with marketing and customer success teams.

Q-World

July 2021 - August 2021

Research Assistant

• Conducted pioneering research in realm of quantum reinforcement learning, with a specific focus on Wigner’s Friend Protocol.
• Awarded with third position in Q-Intern program for impactful nature of research

Languages and Databases

Frameworks

Tools