M.S. Data Science · University of Michigan–Dearborn
Hi, I’m Leela.
I build machine learning systems that turn messy, real-world data into something a person can actually use — a recommendation, a dashboard, a decision.
Grounded in the fundamentals,
curious about the edges.
I recently finished my Master's in Data Science at the University of Michigan–Dearborn, after a background in computer science and a postgraduate program focused on applied ML. Most of what I build lives at the intersection of computer vision, NLP, and cloud infrastructure — the parts of a project that turn a notebook experiment into something that actually runs.
I like problems with a physical or human edge to them: recognizing a face at a door, reading lips from silent video, telling a farmer which crop to plant. The math is the same everywhere, but the constraints — latency, data quality, who's on the other end — change the whole design. That's the part I enjoy most.
Let’s build something great together · Open to full-time Data Science & AI/ML roles
The stack behind the projects
Core & Frameworks
The everyday toolkit for getting from raw data to a trained model.
Machine Learning
Classical ML for tabular, imbalanced, and structured problems.
Data Engineering & Cloud
Getting data pipelines and models to run somewhere other than a laptop.
NLP & GenAI
Language models, retrieval, and topic modeling.
Ten builds, one habit: ship it end to end
Smart Door Recognition
Real-time facial recognition access system using AWS Rekognition and Kinesis video streams, with an IoT trigger for the physical lock.
F1 Concierge Chatbot
A conversational assistant on AWS Lex that answers Formula 1 fan questions, backed by a Lambda/Node.js integration layer.
Lip Reading AI (LipNet)
Reimplementation of the LipNet architecture — a CNN + LSTM pipeline that reads speech directly from silent video frames.
Smart Soil Crop Recommender
A classification model that recommends the best crop for a plot of land from soil composition and climate data.
Auto License Plate Recognition
OpenCV-based pipeline that detects and reads vehicle license plates from video using EasyOCR, tuned for real-world lighting.
Enigma Mental Health Chatbot
A Streamlit-based conversational support tool exploring NLP techniques for empathetic, context-aware responses.
Twitter Stock Analysis (APT)
Applies Arbitrage Pricing Theory to time-series stock data alongside Twitter sentiment to study market movement.
Analysis of the US at UN Debates
NLP-driven analysis of decades of UN General Debate speeches, tracing shifts in US rhetoric and topic emphasis over time.
AI Car Simulation (NEAT)
Self-driving toy cars trained with the NEAT algorithm, evolving neural network topologies through reinforcement-style selection.
ER Visit Power BI Dashboard
An interactive Power BI dashboard analyzing emergency room visit patterns to surface capacity and staffing insights.
Academic foundation
Sep 2024 – Dec 2025
University of Michigan–Dearborn
M.S. in Data Science · GPA 3.78
Jul 2023 – May 2024
Vellore Institute of Technology – Bangalore
Post Graduate Program in Data Science · GPA 3.48
Jul 2019 – May 2023
Vellore Institute of Technology – Amaravati
B.Tech in Computer Science & Engineering · GPA 3.00
Python for Everybody
Responsible AI: Applying AI Principles with Google Cloud
Always learning — more on the way.
Have a problem worth throwing data at? Let’s talk.
Open to full-time data science and ML engineering roles. I usually reply within a day.