17th June 2025

Built this MCP-Powered RAG using Firecrawl and Qdrant! Use Case: Instantly retrieve, summarize, and ground LLM responses in both local documents and live web data for research, knowledge management, or AI assistants.
View on GitHub →

28th May 2025

Taught my last class at Stanford's Code In Place as a Section Leader! Read about my journey here →

Hi, I am

Sudeeksha Vandrangi

I am a

I specialize in machine learning, big data analytics, and AI-driven solutions across finance, retail, and customer analytics. Skilled in developing scalable models and leveraging predictive analytics to optimize decision-making and drive business impact.

Resume
Sudeeksha Vandrangi

Skills

Programming & Databases

Python SQL Scala

Frameworks & Libraries

PyTorch Tensorflow-Keras PyTorch-Forecasting Scikit-Learn Pandas XGBoost NumPy

Big Data & Visualization

PySpark Apache Spark Tableau PowerBI

Data Science & Cloud

Time Series Analysis Demand Forecasting NLP A/B Testing Databricks Azure ML Studio

Experience

Data Scientist Intern

Lennox International, TX

June 2024 - August 2024

  • Designed a multivariate time series forecasting model using a Temporal Fusion Transformer trained on 9M+ data points to predict HVAC schedules for thermostat setpoints.
  • Optimized model predictions and performance using Apache Spark and PyTorch-Forecasting by analyzing processing over 5M+ data points daily on Azure ML Studio and Databricks, increasing forecast accuracy by 10%.
  • Implemented real-time feedback loops via Azure Communication Services, streamlining continuous improvement through email and SMS notifications.
  • Successfully deployed the forecasting model into production, with the project set to go live in Q3 of 2025.
Time Series PyTorch Apache Spark Azure ML Databricks

Technical Trainee Intern

Tata Steel Pvt. Ltd., India

May 2022 - July 2022

  • Decreased defective slab production by 15% by performing root cause analysis using SQL and Python, identifying key process inefficiencies and improving material quality.
  • Built a logistic regression model with 90.43% accuracy to predict defective units and automated monthly defect reporting, reducing reporting time by 40%.
  • Recognized for exceptional performance and offered a pre-placement position to join Tata Steel full-time.
Python SQL Logistic Regression Data Analysis

Projects

MCP Agentic RAG Project

MCP - Powered Agentic RAG with Firecrawl and Qdrant

June 2025

Python FastMCP Qdrant Firecrawl Docker

An extensible Retrieval-Augmented Generation system combining Qdrant vector search and Firecrawl web crawling for fast, semantic retrieval. Powered by FastMCP tool orchestration, it enables intelligent document and web search workflows with modular integration into developer tools like Cursor IDE.

View Project
Modular LLM Orchestration and Feedback Pipeline

Modular LLM Orchestration and Feedback Pipeline

April 2025

Python Streamlit OpenAI Cursor

A local-first orchestration system for managing, auditing, and improving AI-generated qualitative research workflows. It features task-based LLM routing, prompt versioning, and comprehensive logging of all interactions. Integrated Streamlit dashboards enable human-in-the-loop feedback review and full traceability across iterations.

View Project
Data Science Mentor AI Project

Data Science Mentor AI Assistant

March 2025

Python Gradio OpenAI Claude LLM

An AI-powered assistant that helps data scientists grasp complex concepts, debug code, and access documentation with multi-modal support.

View Project
AI Medical ChatBot Project

AI Medical ChatBot

March 2025

Python Streamlit Gemini Pinecone LangChain

A sophisticated medical chatbot providing accurate medical information using Gemini 2.0 Flash and vector search capabilities.

View Project
Earthquake Prediction Project

Earthquake Prediction System

May 2024

Python Random Forest SVC Gradient Boosting Machine Learning

Developed a machine learning system to predict earthquakes using multiple classifiers, achieving high accuracy in seismic event prediction.

View Project
CodeNudge Project

CodeNudge: AI-Powered Technical Interviewer

October 2024

Python Langflow WhisperAPI LLM

Built a functional LLM-driven interviewer prototype in 24 hours, integrating Whisper API for speech-to-text conversion and real-time feedback.

View Project
CLTV Project

CLTV Prediction & Customer Segmentation

April 2024

Python Gradient Boosting Mini Batch K-Means Machine Learning

Developed a comprehensive ML solution for Vahan Bima insurance, achieving 92.53% R² accuracy in CLTV prediction and improving resource allocation efficiency by 15%.

View Project
PII Data Detection Project

PII Data Detection Using NLP

March 2024

Python NLP DistilBERT PyTorch Deep Learning

Developed an NLP-based system using DistilBERT to detect and protect Personally Identifiable Information in educational data.

View Project
Coca-Cola Sales Analysis Project

Coca-Cola Sales Analysis Dashboard

March 2024

Excel Pivot Tables Data Visualization Business Analytics

Created an interactive sales dashboard analyzing Coca-Cola's sales data, identifying key trends and regional performance.

View Project
Hypothesis-Driven Analysis on Credit Card Data

Hypothesis-Driven Analysis on Credit Card Data

March 2024

Python Statistics Hypothesis Testing Logistic Regression Data Analysis

Conducted statistical analysis on credit card customer data using hypothesis testing and logistic regression to identify key factors affecting credit limits and customer attrition.

View Project

Education

Arizona State University

Master of Science in Data Science, Analytics, and Engineering

August 2023 - May 2025

GPA: 4.0/4.0

National Institute of Technology (NIT), Rourkela

Bachelor of Technology in Metallurgical and Materials Engineering

July 2019 - June 2023

GPA: 8.91/10.0

Certifications

NVIDIA Logo

Introduction to Transformer-Based Natural Language Processing

NVIDIA - January 2025

View Credential
University of Michigan Logo

Using Python to Access Web Data

University of Michigan - June 2020

View Credential

Contact

Feel free to reach out to me for any questions or opportunities!

Email Me ✍️