cv

ML Researcher

Gardos Lab @ BU CDS

Working on a multi-modal foundation model for analyzing millions of Herbaria samples and educational AI tutors. Advised by Dr. Thomas Gardos.

• Language Models
• Computer Vision
• AI Agents

MS Researcher

Image and Video Computing Group @ BU CS

I work on video-based human-computer interfaces and assistive technology. I lead the redesign of CameraMouse, a desktop application that enables people with severe motor impairments to use a computer with their head movement. I work on advanced face-tracking and gesture recognition capabilities to improve user interaction, developed a testing GUI for analyzing mouse movement efficiency, and created the testing routine that included both target selection and typical daily tasks. Advised by Dr. Margrit Betke.

• Human-computer interaction
• Facial recognition
• Assistive technology

Engineering Intern

Software Application and Innovation Lab

Network graph visualization of the ASL lexicon; Data pipeline and dashboard for advertisement, marketing and internal analytics; Debugging Android API and OS issues.

• Linguistics research
• Data Visualization
• Android and desktop applications

Data Engineer

Smart Solutions Group

Automation of scraping e-tender activity in the banking industry; Collection and analysis of social media posts; Using SQL to fetch data relevant to analysis of trends and fraudululent marketplace system activity.

• Social Media analytics
• Web Scraping
• Data Visualization

MS

Boston University

Data Science

• Deep Learning
• Computational Statistics
• AI Ethics
• Time Series

BA

Boston University

Computer Science

A Head-Based Mouse-Control System for People with Severe Disabilities

Farid Karimli, Hao Yu, Manny Akosah, and Margrit Betke, Wenxin Feng

We propose a complete AI-based redevelopment of the CameraMouse interface that includes automated facial feature detection and new ways to map facial feature positions to mouse coordinates and facial feature movements to mouse clicks. In addition to the traditional dwell-time mechanism, users can select items on the screen by opening their mouth or raising their eye brows.

AirMouse

Control your mouse by finger pointing and hand gestures.

• Computer Vision
• Desktop Software
• Open Source

VideoASL

Using masked autoencoders and space-time attention to classify extended American Sign Language in video.

• Video classification
• Space-time Attention

FGK.ai

Attemping to build a chatbot that mimics my style of texting and writing. Building using Chainlit, LangChain and OpenAI.

• AI Agent
• AI Assistant