Hi there, I’m

Shaurya

Hi there

My name is Shaurya, I’m an AI product builder and data scientist, currently pursuing my Master’s at Carnegie Mellon University. I have a passion for building intelligent, useful systems, from generative AI tools and RAG pipelines to real-world agentic applications.

Outside of tech, I’m deeply curious about human behavior, storytelling, and systems thinking, and I strive to create solutions that blend utility with user impact.

Below are a few projects I’ve built to learn, solve problems, and push boundaries, both individually and with teams.

Deep-Dive?

Education



Carnegie Mellon University

Master of Information Systems Management

Business Intelligence and Data Analytics, 2024–2025


Chandigarh University

Bachelor of Engineering – Computer Science

AI & ML Specialization, 2018–2022

Experience



Research Assistant – Carnegie Mellon University

May 2025 – Present

Working under faculty guidance on building a multi-agent RAG system for navigating legal and policy documents. Responsible for implementing document chunking, embedding generation, vector storage, and conversational agent workflows using Microsoft AutoGen.


Computing Consultant – Carnegie Mellon University

August 2025 – Present

February 2025 – May 2025

Assist with the implementation of AI-driven tools to improve efficiency and user experience. Provide consulting and guidance on integrating AI solutions into computing workflows, ensuring alignment with student and faculty needs. Collaborate with Heinz Computing staff to maintain and enhance computing services.


Data Analyst – Carnegie Mellon University

February 2025 – May 2025

Conducted exploratory and statistical analysis on faculty survey data to uncover patterns in course feedback, TA performance, and student experience.


Software Engineer – YMGrad

2022 – 2024

Contributed to the backend architecture of YMGrad, a platform supporting international students. Focused on optimizing database queries, managing large-scale user data, and ensuring secure API performance.

Projects



EU Navigator


Multiagentic RAG Personal Learning Portal

EU Navigator is a sophisticated Multiagentic Retrieval-Augmented Generation (RAG) system designed as a Personal Learning Portal for exploring European Union tech, data, and AI legislation. The system integrates multiple specialized AI agents for planning, retrieval, synthesis, and review—enabling evidence-based legal analysis with verified citations.

Built using LangChain, FAISS, and Llama 3.1, it preserves EU legal structure through section-based chunking and semantic search. A Streamlit-based learning interface supports modular exploration, progress tracking, and reflection journaling. Evaluation was performed with the RAGAS framework to measure answer faithfulness and relevancy.

View Project


UCI Air Quality Prediction with MLFlow

Developed a complete machine learning pipeline for predicting air quality using the UCI dataset. The project included experiment tracking with MLFlow, real-time data ingestion through Kafka, and a deployed prediction API containerized with Docker.

Built advanced monitoring dashboards with Evidently AI to detect data drift and performance degradation. Focused on reproducibility, modular ML design, and robust model serving for continuous deployment and evaluation.

View Project


Question Answering System using BERT & Azure ML

Built a fine-tuned BERT-based question-answering system capable of matching user queries with relevant answer sentences from the SQuAD dataset. The system combined semantic similarity scoring, sentence embeddings, and Jaccard/TF-IDF baselines for benchmarking.

Deployed the model using Azure Machine Learning, optimizing the scoring pipeline for latency and accuracy. This project demonstrated full-stack ML deployment, including model fine-tuning, packaging, serving, and real-time inference.

View Project


Fairness Audit Playbook


Responsible AI Evaluation Framework

Created a comprehensive Fairness Audit Playbook that standardizes fairness evaluation for AI systems across domains. The framework integrates bias detection, intersectional fairness analysis, and historical context assessment to promote responsible AI adoption.

The playbook includes fairness definition selection, bias source identification, validation workflows, and implementation guidelines designed for engineering teams. Presented with case studies and visualization modules for explainability.

View Report


and more!

Contact

Reach out on LinkedIn
or email me at i.shauryagulati@gmail.com

Booting...

Starting...