Color Scheme

Machine Learning Engineer

Tiyasa Saha

I’m a data scientist with an MS in Data Science from the University of Massachusetts Dartmouth, passionate about turning data into impactful solutions. My experience spans building healthcare-focused AI systems, from breast cancer treatment prediction and brain tumor detection using MRI scans to LLM-powered chatbots. I work extensively with Python, SQL, TensorFlow, and Keras, and I have hands-on experience deploying models into production with frameworks like Flask. Beyond the technical side, I value clear communication and collaboration, and I’m driven by a vision to apply AI in ways that improve healthcare outcomes and create meaningful impact.

Tiyasa Saha.

Projects

Brain Tumor Classification (MRI) using TensorFlow

Developed an end-to-end brain tumor classification system that predicts tumor type from MRI scans using a trained deep learning model. The solution includes a React.js frontend for image upload and result visualization, and a FastAPI backend that serves a TensorFlow model for real-time inference.

  • React.js
  • FastAPI
  • TensorFlow
  • Keras
  • Python
  • Deep Learning
View on GitHub
Brain Tumor Classification project screenshot.

BrCaBot - Breast Cancer Information Chatbot

Built an LLM-based breast cancer chatbot using LangChain, Pinecone, and RAG with OpenAI and Hugging Face embeddings. Deployed using Flask with a custom dark-mode chat UI for an intuitive and accessible user experience.

  • LangChain
  • Large Language Models (LLM)
  • Pinecone
  • Natural Language Processing (NLP)
  • Retrieval-Augmented Generation (RAG)
  • CI/CD
View on GitHub
BrCaBot project screenshot.

Research Article LLM Tool

An LLM-powered research assistant that uses LangChain, OpenAI, and FAISS to answer queries based on articles entered via a Streamlit interface.

  • Streamlit
  • Retrieval-Augmented Generation (RAG)
  • FAISS
View on GitHub
Research Article LLM Tool project screenshot.

BRCA (Breast Invasive Carcinoma) Treatment Prediction System

Developed a machine learning–based treatment prediction system that analyzes patient data to suggest personalized treatment options. The model was deployed using Flask and integrated into an interactive web application, enabling real-time predictions and a user-friendly interface for healthcare practitioners and patients

  • Flask
  • Postman API
  • Jupyter
View on GitHub
BRCA Treatment Prediction project screenshot.

Data Visualization of Mass Shootings in USA

Designed an interactive webpage visualizing U.S. mass shooting trends (2014-2022) with HTML, CSS, and d3.js.

  • NetworkX
  • D3.js
  • Web Design
View Project
Data Visualization of Mass Shootings project screenshot.