Hey, I'm Mohsin
I am a builder, I create software, engage in entrepreneurial ventures, lead communities, and challenge myself through fitness.
Where I've Worked
UC Berkeley College of Engineering
Conducted research on an exoskeleton project. Co-developed a machine learning algorithm using EEG brainwave data to detect user intent and improve motion control. Also built real-time signal classification models using Python, scikit-learn, TensorFlow, and MNE-Python, supporting neurotech applications for mobility and accessibility.
InterestingSoup
Engineered and deployed scalable RESTful APIs using Python and Flask, reducing response time by 30% and supporting a 20% increase in concurrent users and implemented full-stack development. Partnered with cross-functional teams to build and integrate a real-time analytics dashboard using React and Chart.js, enhancing data visibility and enabling faster decision-making.
What I've Built
NeuroFocus
NeuroFocus is a machine learning project that classifies cognitive attention states—Focused, Distracted, and Overstimulated—by analyzing EEG data from the DEAP dataset. I built a complete ML pipeline that extracts key features from EEG frequency bands (Theta, Alpha, Beta, Gamma) and applies models like SVM, LSTM, and CNN to achieve up to 90% classification accuracy. The system is designed with real-world applications in mind, such as mental focus tracking, neurofeedback, and brain-computer interface (BCI) systems. This project highlights the intersection of neuroscience and AI, showing how neural signal processing can be used for accurate, real-time cognitive state monitoring.
EvoScope
EvoScope is an advanced AI trading system that combines cutting-edge machine learning techniques for autonomous financial trading. The system features a sophisticated architecture with LSTM models for time-series prediction and reinforcement learning agents for intelligent decision-making. Built with real-time data processing capabilities, EvoScope initializes multiple AI models simultaneously to analyze market patterns, execute trades, and optimize portfolio performance. The platform showcases the evolution of algorithmic trading through deep learning, featuring a sleek interface that visualizes model initialization, trading decisions, and performance metrics in real-time.
LLM Sensitivity Evaluation in Political Contexts
This project explores how large language models (LLMs) respond to politically sensitive prompts, leveraging techniques in Natural Language Processing (NLP) and Automatic Prompt Evaluation (APE). The pipeline systematically probes LLMs with ideologically varied inputs and quantifies their response consistency and sensitivity. It offers a practical framework for analyzing model alignment, political bias, and linguistic stability—critical challenges in modern NLP. The insights contribute to building more robust, transparent, and socially-aware AI systems.
Let's Connect
I'm always interested in hearing about new projects and opportunities. Whether you have a question or just want to say hi, I'll try my best to get back to you!
WHO I AM
I grew up in the San Francisco Bay Area. I completed my undergrad at UC San Diego in Cognitive Science (Machine Learning) and Computer Science. I love building and scaling AI projects.
I am a builder. I create software, engage in entrepreneurial ventures, lead communities, and challenge myself through fitness. Whether it's coding or CrossFit, I appreciate the struggle and tend to fall in love with the process.
Apart from work, I enjoy staying active through soccer, tennis, biking, and running. I'm also passionate about traveling to new places and winding down with a good book whenever I can. (check out my Strava if you're curious)!
PHOTO GALLERY
MY LIBRARY
