Hello, I'm
Atul Kumar
Machine Learning & Data Enthusiast
Building intelligent algorithms, automating data pipelines, and transforming unstructured data into actionable insights.
Building intelligent algorithms, automating data pipelines, and transforming unstructured data into actionable insights.
Engineering student at Birla Institute of Technology, Mesra (B.Tech Chemical, 2023–Present). Passionate about technology-driven problem-solving and machine learning.
My toolkit includes Python, PyTorch, Scikit-learn, and Web Technologies. I thrive on processing datasets, designing RAG systems, and building AI-driven utilities to deliver innovative digital solutions.
Pursuing B.Tech with an emphasis on data-driven chemical modeling, computational physics, and process simulation. Actively applying machine learning algorithms to chemical analytics and predictive optimization workflows.
Completed secondary school with deep focus on Physics, Chemistry, Mathematics, and Computer Science fundamentals, establishing a core engineering base.
Developed an AI-powered Stock Research Assistant integrating 4+ analysis modules for technical, fundamental, and sentiment analysis. Built a FastAPI backend with 10+ REST API endpoints for automated stock data processing, implementing LLM-based summarization of 8+ financial indicators that reduced manual research time by 70%.
Built an enterprise RAG assistant using FastAPI and LangChain with a 5-stage retrieval pipeline. Implemented semantic search via vector embeddings to improve response relevance by 40% and reduce latency by 35% using an end-to-end PDF parsing and chunking pipeline with 100% citation-backed answers.
Analyzed financial market behaviors under varying conditions by detecting market regimes (Bull, Bear, Sideways) using unsupervised machine learning. Engineered rolling momentum and volatility features, applied K-Means clustering on NIFTY 50 index data, and evaluated simple trading strategy performance and stability transitions.
Developed a financial analytics pipeline that ranks stocks based on risk-adjusted performance rather than raw returns. Computed key metrics including volatility, maximum drawdown, and Sharpe ratios using normalized composite scores, and created an interactive visualization framework using Streamlit.
Developed a Python-based automation tool to send personalized internship/job application emails using structured CSV data and secure SMTP authentication. Implemented custom data sanitization and validation pipelines to process recipient lists, integrated built-in rate limiting to optimize delivery schedules and minimize spam flags, and established secure credential management.
Developed a YOLOv8-based instance segmentation pipeline for detecting dental caries from 750+ radiographic images. Trained on 768×768 images using AdamW optimizer with cosine learning-rate scheduling, applying augmentation techniques to increase dataset diversity by 4× and achieving 22.6% mask recall.
Built an AI-powered no-code Chrome extension generator utilizing React, Node.js, and Groq AI. Integrated the Monaco Editor for code visualization and custom configuration editing, automated the packaging of Manifest V3 compliant files into download-ready ZIP files, and established a secure backend system.
Completed tasks simulating real-world analytics, dashboard design, and translating complex metrics into client insights.
View CertificateComprehensive bootcamp covering statistical operations, DBMS, SQL programming, and advanced Tableau dashboard creation.
View Certificate