NYU MS Data Science Student | Deep Learning Enthusiast
I believe AI should democratize innovation, not gatekeep it. My mission? To architect intelligent systems that transform decision-making and build solutions that empower people to focus on creative problem-solving, breakthrough innovation, and high-impact work.
I'm a computational problem-solver, orchestrating the convergence of statistical modeling, ML architecture, and process automation to engineer intelligent systems with real-world impact. With extensive experience in data science, I've designed enterprise-scale ML solutions, optimized mission-critical workflows, and pioneered cutting-edge NLP frameworks and predictive credit scoring models.
From deep neural networks to distributed data engineering pipelines, I specialize in transforming complex technical concepts into accessible, efficient solutions.
Technologies I've worked with and mastered over the years
Python
PyTorch
TensorFlow
Scikit-Learn
NumPy
Pandas
R
Hugging Face
Seaborn
Git
Docker
AWS
PostgreSQL
LaTeX
Grafana
Sentry
Datadog
Jira
Notion
Postman
HTML
CSS
My journey through the years
Academic excellence and competitive achievements over the years
Test of English as a Foreign Language
Graduate Management Admission Test
Graduate Record Examinations
Madhava Mathematics Competition
Madhava Mathematics Competition
Indian Institutes of Technology Joint Entrance Examination
Joint Entrance Examination Main
Indian National Mathematics Olympiad
Kishore Vaigyanik Protsahan Yojana
National Talent Search Examination
Articles and case studies from my recent work
Kaggle BirdCLEF+ 2026: multi-label classification of 650+ wildlife species from continuous Pantanal soundscapes using EfficientNet SED + iterative pseudo-labeling + dialect clustering.
Geometric analysis of adversarial manipulation in human cognition: TinyRNN phase portraits reveal that adversaries work by inverting cognitive attractors, not nudging individual decisions.
Multimodal AI agent that classifies street-level infrastructure issues from photos, voice, or text and submits structured 311 complaints in 17 languages via SMS, email, or call.
Extending the Latte Latent Diffusion Transformer with physics-informed training objectives to generate text-conditioned videos that respect real-world physical constraints.
Advanced NLP pipeline that combines regex patterns and LLM capabilities to extract and label financial entities from SMS data, enabling comprehensive user behavior analysis.
NLP pipeline that generates optimized regex patterns from unstructured text, enhancing entity extraction with 93% accuracy.
Self-organizing system using FastText embeddings and clustering to categorize merchants with 87% accuracy.
Hierarchical deep learning framework achieving 91% accuracy across 27 categories with reinforcement learning optimization.
Financial risk assessment engine using gradient-boosted trees to predict default probability with 89% precision.
Exploring creativity through various mediums
Interactive AI-powered games and applications showcasing machine learning in action.
Explore my collection of wildlife and nature photography, capturing moments of beauty in the natural world.
Discover my origami creations, from simple geometric shapes to complex artistic designs.
Explore my artistic creations and design work, showcasing creativity across various mediums and styles.
A visual journey through photography, origami, and art
