Professional Profile and Technical Projects of Shaurya Rawat

Candidate: Shaurya Rawat, B.Tech AI/ML student at GGSIPU – USAR, Delhi ( $2025$ – $2029$ ).
Academic Standing: Current CGPA of $8.64$ .
Core Focus: Understanding fundamental mechanisms of Machine Learning, reproduction of research papers, and open-source contribution via GSSoC '26.

ML / Deep Learning: PyTorch, scikit-learn, XGBoost, CNNs, ResNets, RNNs, LSTMs, SVM, Random Forest, and PCA.
GenAI & Agents: LangChain, LangGraph, LLM APIs, and Prompt Engineering.
Deployment & Tools: FastAPI, Docker, REST APIs, Pydantic v2, Streamlit, and Git/GitHub.
Foundations: Proficiency in Linear Algebra, Probability, Statistics, and Optimization.

ResNet Degradation Reproduction Study: Verified He et al. (CVPR 2016) using CIFAR-10.
- Demonstrated optimization failure: CNN-56 ( $58.94\%$ ) vs. CNN-20 ( $82.06\%$ ).
- Validated residual fix: ResNet-32 outperformed ResNet-20 by $+3.20\%$ .
Insurance Premium Prediction API: Built a REST API using FastAPI and Pydantic v2 for premium predictions, containerized with Docker.
Applied ML & Deep Learning Repositories: Over $15+$ projects implementing algorithms like Linear Regression, Logistic Regression, SVM, and PCA from scratch in NumPy to understand library internals.

AI Agents Contributor (GSSoC '26): Building workflows with Fetch.ai Innovation Lab.
NagarAI Grand Finalist: Led a $4$ -person team to create an NLP pipeline for prioritizing citizen complaints; ranked among $1,000+$ participants.
VibeClash 2025: Team Leader for a group that placed in the Top $20$ out of $192$ teams.

Certifications: Deep Learning (Mike X Cohen/Udemy), Google Machine Learning Crash Course, and IBM Introduction to Large Language Models.
MVPP Scholar: Ranked $310$ in Delhi State (top $1\text{--}2\%$ ) in $2021$ .
SilverZone Science Gold Medal: School Rank $1$ and National Olympiad Rank $107$ in $2021$ .