Hector Nevarez
Software Engineer
(619) 852-7292
Eglen.neva@gmail.com
Website: hectorenevarez.github.io
Education
Bachelor of Science in Computer Science August 2018 – December 2021 (Expected)
San Diego State University
Cumulative GPA: 3.52
Major GPA: 3.50
Skills
Programming: C/C++, Python, Linux, JavaScript, HTML/CSS
Tools: Git/Github, TensorFlow, scikit-learn, pandas, Jupyter Notebook, CMake, OpenCV, GStreamer
Experience
Software Engineer Intern February 2021 – Present
ModalAI
• Integrated depth from stereo from forward and rear facing camera pairs and left and right facing ultrasonic range
finders, on a drone, for vehicle obstacle avoidance
• Created drivers for ultrasonic range finding sensors by developing an IO layer that interface with an IO expander
via I2C and utilized the IO layer to program sensor registers
• Ported a visual inertial odometry application from a 32-bit embedded device to a 64-bit embedded device by
developing a camera server utilizing GStreamer, using named pipes to communicate data between processes, and
synced camera and IMU timestamps
• Designed a system monitor tool for an embedded device that runs as a background process with an optional client
interface that reports CPU, GPU, and memory usage
Chief Technology Officer | Workshops January 2020 – August 2021
San Diego State University A.I. Club
• Created AI related workshops that introduced club members to fundamental topics such as Python, Data Science,
Computer vision, and Neural Networks and presented material weekly to over 50 club members
• Introduced and managed project division which promotes the involvement of club members working on artificial
intelligence related projects with advising from club officers
Projects
COVID-19 Risk Detection | GitHub
• Created data collection software, using YOLOv4 for person detection, to develop a face mask dataset with over
3,000 images from over 20 hours of collected CCTV footage
• Developed a computer vision model with TensorFlow that classifies whether an individual is wearing a face mask
with 91% accuracy through transfer learning
• Designed a multi-object tracking algorithm that re-identifies all people each frame, by measuring difference in
Euclidean distance, in order to speed up software
