Soumi Das

Profile

I am a dual-degree student pursuing a B.Tech in Computer Science and a BS in Electronic Systems. Through personal projects, I have gained experience in embedded systems, electronics, and machine learning. I am an inquisitive learner with strong problem-solving skills, eager to take on challenges and contribute effectively.

Projects

Home Security System

Arduino Nano, C/C++, RFID, I²C, SPI, sensorsDec. 2025

• Arduino Nano-based, multi-module home security system.
• Built RFID-based authorization using an MFRC522 module over SPI.
• Displayed real-time system status on an LCD1602 screen over I²C, with a buzzer alert for intrusions.
• Implemented a relay-driven solenoid door lock, laser tripwire window security, and flame-sensor fire safety with emergency door release.

Repository Video Demo

Interactive LED Animation System

Arduino Uno, Embedded C/C++, MAX7219, SPINov. 2025

• Designed an interactive display system to render custom animations on an 8×8 LED dot matrix.
• Integrated a TTP223 capacitive touch sensor as the user interface.
• Programmed dynamic cycles so users can rotate through three distinct animations with a touch.

Repository Video Demo

PRISM: Severity Prediction & Response Strategies

Python, Supervised Learning, Linear Regression

• Severity Prediction and Optimizing Response Strategies Using AI and Machine Learning.
• Analyzed casualties, economic losses, and emergency-response data from the Global Disaster Response Analysis dataset.
• Used supervised linear regression to predict a severity index for global disasters.
• Achieved 97.51% accuracy and an RMSE of 0.4432 to support resource allocation and proactive disaster recovery.

Repository Poster Paper

Global Disaster Severity Prediction Model

Python, scikit-learn, NumPy, PandasDec. 2025 – Present

• Predicts disaster severity using factors including casualties and economic loss.
• Applied linear regression for prediction and KNN to visualize results.
• Evaluated tolerance and accuracy using MAE, MSE, and RMSE metrics.

Repository

Fitsphere Model

FastAPI, Python, scikit-learn, Joblib, UvicornDec. 2025 – Present

• Recommends personalized exercises by analyzing user health metrics (BMI, height, calorie intake).
• Created an app feature that recommends personalized exercises by analyzing user health metrics, including BMI, height, and calorie intake.
• Used ensemble learning to train a Random Forest classifier for recommendations.
• Built a FastAPI service to handle user data and deliver exercise suggestions.

Repository

Education

Technical Skills

Languages

C/C++, Python, SQL (MySQL), JavaScript, HTML, CSS, PHP

Frameworks

React, Node.js, FastAPI, Arduino

Libraries

Pandas, NumPy, Matplotlib, scikit-learn, Joblib, Uvicorn

Developer tools

Git, VS Code, PyCharm, Google Colab