Open to AI/ML internship opportunities. I build RAG systems, predictive ML pipelines, and deep learning models, and ship them as working apps.

• 🎓 B.Tech in Artificial Intelligence & Data Engineering, IIIT Kota (2024–2028) | CGPA: 7.88
• 🧠 Focus: Generative AI & RAG, voice-enabled LLM systems, predictive ML, and FastAPI backends
• 💻 425+ DSA problems solved on LeetCode, CodeChef, and Codeforces (C++)
• 🎯 Goal: turn strong ML fundamentals into deployable, tested projects

Domain-specific RAG, OCR & Voice AI

• Built a grounded RAG system over land-acquisition records using FastAPI, LangChain, ChromaDB, and Ollama.
• Ingests digital and scanned documents (PyMuPDF + Tesseract OCR) and generates standardized reports automatically.
• Integrated Sarvam AI for voice interaction: speech-to-text queries and spoken LLM responses.

LangChain • FastAPI • ChromaDB • Ollama • Tesseract OCR • PyMuPDF • Sarvam AI

💻 Code • 🎥 Demo

Full-stack industrial ML with a React dashboard

• Trained a Random Forest machine-failure classifier on the UCI AI4I 2020 dataset (10,000 records) using leakage-aware feature selection and stratified splits.
• Reached 83.67% precision, 70.09% F1, and 78.17% average precision on the held-out test set at a tuned threshold of 0.40.
• Deployed a FastAPI inference service and a React + TypeScript dashboard with input validation, automated tests, and model-artifact integrity checks.

Python • Scikit-Learn • FastAPI • React • TypeScript

💻 Code

Human vs. AI-generated speech classification

• Built a deep learning pipeline to detect AI-generated speech across 5 languages: English, Hindi, Tamil, Telugu, and Malayalam.
• Extracted MFCC spectral features from preprocessed audio to train and evaluate the classifiers.

TensorFlow • Audio DSP • MFCC • Python

💻 Code

Geospatial transfer learning on EuroSAT

• Built a land-use classification pipeline with ResNet-50 using TensorFlow and Keras.
• Two-stage transfer learning: 95% accuracy with frozen pretrained layers, then 98% validation accuracy after fine-tuning the final 30 layers.

TensorFlow • Keras • ResNet-50 • Computer Vision

💻 Code

• Optimized C++ solutions covering DP, Graphs, Trees, Two-Pointers, and Bit Manipulation.
• Built through LeetCode, CodeChef, Codeforces, and the Striver A2Z Sheet.

💻 Code

Languages

AI / ML

Backend, Data & Tools

• ☁️ Oracle Agentic AI Certified Foundations Associate
• 🥇 CPS Olympiad: All India Rank 258, State Rank 86
• ⚡ Amazon ML Summer School 2026: Qualified for the national selection test (top 31,820 of 134,420 registrations)
• 🚀 ET AI Hackathon 2.0: Qualified for the online round (top 9,629 of 60,974 registrations)
• 🎓 NSP Scholar
• 📜 Mastering Data Structures & Algorithms using C and C++ and Learn C++ Programming: Deep Dive in C++ (Udemy)

"Learning is useful. Building is what proves it."