📧 harshrofff@gmail.com | 🔗 LinkedIn | 🔗 Website
I am a data science graduate student and a research assistant at UMBC, where I develop prototypes on robots from ClearPath Robotics, implement LIDAR and camera interface, and build indoor and outdoor maps. I am also developing teleportable control schemes using hand gestures and hollow lens for Boston Dynamics' Spot robots. I'm utilizing my expertise in Python, AI, and ML to create innovative solutions to complex problems and drive value and insights through data analysis.
I have a bachelor's degree in Electronics and Communication Engineering from Vishwakarma Government Engineering College, GTU. I have also completed several certifications in Python, AI, and ML from IBM. I have published a paper on mosquito recognition using ML and Edge Impulse on a tiny embedded system. I am passionate about learning new technologies, collaborating with diverse teams, and contributing to the advancement of science and engineering.
Languages: C/C++, Embedded C, Python, R, SQL, HTML, CSS
Tools & Frameworks: Django, TensorFlow-Keras, PyTorch, Seaborn, Streamlit, Matplotlib, SciPy, NLTK
Development Boards: Arduino Uno, Nano 33 BLE, RaspberryPi, Jetson Nano, STM32 Boards
Master of Data Science
University of Maryland, Baltimore County (UMBC)
GPA: 3.9/4.0
Aug 2022 - May 2024
Relevant Coursework: NLP, Data Analysis & Visualization, Data Management, Big Data Processing, Financial Data Science, Machine Learning, Artificial Intelligence
Bachelor of Engineering in Electronics & Communication Engineering
Gujarat Technological University
GPA: 3.4/4.0
Aug 2019 - May 2022
Relevant Coursework: Python, Artificial Intelligence, Internet of Things
Graduate Research Assistant
Center for Real-time Distributed Sensing and Autonomy, UMBC
Mar 2023 – Present
- Developing hand gesture based control for Boston Dynamics' Spot robots for teleoperation using Microsoft HoloLens 2.
- Developing prototypes on robots from ClearPath Robotics like Jackal and Husky.
- Implementing LIDAR and interfacing Axis camera on Jackal and building indoor and outdoor map.
Data Science Tech Intern
The Conservation Fund, Freshwater Institute, WV, USA
Aug 2023 - Dec 2023
- Implemented machine-vision algorithms on Raspberry Pi to streamline the traditional salmon coloration process by 80% efficiency.
- Developed a GUI in python to efficiently run the algorithm and process captured images.
- Designed a 3D printed case to encase an LCD screen, Raspberry PI 4, and HQ camera.
Electronics & Machine Learning Intern
WeHear Innovations PVT LTD, Ahmedabad, India
Jan 2022 - Jun 2022
- Developed prototypes for both bone-conduction earphones and wristband to assist hearing impaired people.
- Integrated an ML model with 85% accuracy for speaker recognition with wristband and earphones to facilitate mobility of people with disability.
- Designed Printed Circuit Boards (PCBs) for the second-generation bone-conduction earphones and wristbands to improve performances, results, and accessibility by 57% by introducing touch sensitive buttons, more battery power and new design.
Embedded Systems Engineer Intern
Episodic Labs PVT LTD, Ahmedabad, India
Feb 2021 - Aug 2021
- Integrated 2 camera modules with a RaspberryPi 4 & Compute Module I/O board to achieve best shot with proper focus and framerates.
- Engineered motion-enabled sensors and sports action cameras to automatically trigger the camera and capture 10 seconds of motion while playing a stroke to track and evaluate batsmen’s performance.
Smart Dresser - GitHub
- It is an AI-powered outfit selector designed to make shopping experience faster, easier, and safer.
- Integrated different technologies, including machine learning, image processing, and body posture detection, to provide personalised outfit recommendations based on user preferences and body type.
Svadeshi - GitHub
- Gathered the dataset from two ancient scriptures namely Shrimad Bhagwat Gita and Ramayana to train the model.
- English-to-Sanskrit and vice versa translation tool with 79.8% accuracy built using NLP to preserve the Sanskrit language because Sanskrit speakers are fewer and further between these days.
Dr Tongue - GitHub
- Web application, ranked 4th at a state level Ideathon, designed for tongue diagnosis based on ML model trained using python and integrated with a website.
- Neural network model achieved training accuracy of 80% and testing accuracy of 76% with a small dataset.
WhatsApp Chat Analyzer - GitHub
- Developed a web-app in Python with the help of Streamlit to analyze and visualize WhatsApp chats.
- Displayed graphs of most active users, most used words, most used emojis, and activity time of each user.
Cyclists’ Dashboard - GitHub
- Dashboard coded in Python with Dash module to enable cyclists with performance analysis.
- Coded to help cyclists to analyze main three aspects of the rides namely, average, and maximum speed, distance, and elevation to improve cycling skills.
- It also displays a map of the path taken while riding and analyses of the rides on a monthly and annual basis.
- Published a research paper on “Mosquito Species Identification from its Wing Beats on a TinyML Device” with 17 other papers from 10 countries at Kaleidoscope 2021 conference hosted by the International Telecommunication Union (ITU) IEEE Explore.
- Ranked 1st at an international hackathon organized by the ITU and The Telecommunications and Digital Government Regulatory Authority of United Arab Emirates (TDRA) News Article.
- Presented a talk to 10 teams and guest speakers on “Design of a Python-based Dashboard to Assist Athletes in Analyzing and Enhancing Performances” at the SciPy India 2021 conference YouTube Video.
- Featured in a renowned newspaper, Ahmedabad Mirror for the project Smart Dresser: an AI based Outfit Selector.