Name: Hardik Prajapati
Profile: Machine Learning | Computer Vision | GCP certified
Email: hp.6318@gmail.com
Phone: +1 (213) 414-8703
Skill
PythonAbout me
I am a Machine Learning and Computer Vision Ph.D. candidate at UCSB, where I focus on advancing the fields of video scene graph generation and agent trajectory anomaly detection. My research leverages transformer architectures and graph-based methods to uncover novel insights in dynamic and complex environments. I am honored to work under the guidance of Prof. B.S. Manjunath.
As an AI Engineer at Analytos, I delivered impactful solutions across diverse domains. From designing sales forecasting pipelines that enhanced decision-making by 20% to optimizing vehicle routing with reinforcement learning for a 12% efficiency gain, I have consistently translated cutting-edge algorithms into real-world impact. My collaborative efforts as a Graduate researcher at Media Communications Lab further refined my expertise, culminating in lightweight 3D point cloud classification models.
As a Google Cloud Certified Professional Machine Learning Engineer, I bring a strong foundation in building and deploying scalable ML pipelines on cloud platforms. My technical toolkit includes expertise in Python, PyTorch, TensorFlow, and Google Cloud’s VertexAI, among others. Complementing my technical acumen is a passion for academic rigor, reflected in a perfect 4.0 GPA during my Master's at USC and recognition with the Outstanding Academic Achievement Award.
Beyond research and development, I thrive in collaborative environments. My leadership roles—whether managing a team of 43 at Unilever or coordinating club events as Vice President—highlight my commitment to fostering teamwork and innovation.
With a drive for impactful AI innovation and a commitment to excellence, I am always open to collaborations and discussions. Let’s connect and shape the future of AI together!
Education
B.TECH - Instrumentation & Control Automation
Nirma University - India
July,2014 - May,2018
CGPA: 8.34/10.0 (3.656/4.0)
MS - Electrical & Computer Engineering
University of Southern California - USA
Jan,2021 - Dec,2022
CGPA: 4.0/4.0
Ph.D. - Electrical & Computer Engineering
University of California, Santa Barbara - USA
Sept,2024 - May,2028 (Expected)
Work Experience
Junior AI Engineer
Analytos
Jan,2023 - June,2024
Developed AI-driven solutions, including a sales forecasting pipeline and a reinforcement learning-based vehicle routing optimization system, achieving measurable improvements in decision-making and operational efficiency
Supervisor: Sunil Ranka
Graduate Researcher
Media Communications Lab
May,2022 - September, 2022
Designed and implemented a pose invariant light-weight model for 3d Point cloud object classification. Achieved 84% accuracy with inference time less than 0.5sec and model size of 900kb.
Supervisor: C.-C. Jay Kuo
Course Mentor
USC-Machine Learning
Jan,2022 - Dec,2022
Coached batch of 110 students for Machine Learning concepts and fundamentals. Designed homework and exam problems, delivered solutions and assisted professor with classroom logistics.
Supervisor: Keith Jenkins
Engineer
G6 SuperHomes
Sep,2019 - April,2020
As an early engineer, spearheaded the design and development of a pioneering prototype switchboard (the 'super-switch'), enabling users to effortlessly operate home appliances via touch, mobile application, and voice control, revolutionizing accessibility.
Manager: Sarvam Miyani
Manufacturing Engineer
Unilever-India
July,2018 - Aug,2019
Significantly improved efficiency and savings through automation and data-driven decision-making in addition to supervising a team of 43 shop-floor employees.
Supervisors: Deepesh Bisht, Swati Kumari
Publications
S3I-PointHop: SO(3)-Invariant PointHop for 3D Point Cloud Classification
Conference: IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP), 2023
It assigns a class label to a point cloud scan, whose points are expressed in an arbitrary coordinate system. This is achieved through the derivation of invariant representations by leveraging principal components, rotation invariant local/global features, and point-based eigen features.
A tiny machine learning model for point cloud object classification
Journal: APSIPA Transactions on Signal and Information Processing
A machine learning model which can be deployed in mobile and edge devices for point cloud object classification. It has a model size of 64K parameters. It demands 2.3M floating-point operations (FLOPs) to classify a ModelNet40 object of 1024 down-sampled points.
Professor, former HOD - IC Department, Nirma University
Hardik has always been an excellent student having secured a Rank in the Top 10. During the 4 years, I have known I have seen him have clear thoughts and also stay calm headed. He is always well prepared and completes his work well in advance. He puts in his sincere efforts to accomplish any task he wishes to accomplish.
Senior Research Engineer - Tencent
I had the privilege of collaborating with Hardik on a challenging research project from May to October 2022 during his tenure as an intern in my lab at USC. Hardik stood out as the top candidate during the interview process for the internship, where he presented insightful ideas to improve existing literature. His quick adaptation to our lab's culture and immediate contributions to our 3D point cloud analysis project were impressive. Hardik's proactive engagement during our weekly meetings and frequent sync-ups demonstrated his strong research aptitude and work ethic. Together, we co-authored two papers accepted at prestigious conferences like IEEE ICASSP. Hardik excelled in tackling complex and open-ended problems. Beyond his professional strengths, Hardik is approachable and personable, qualities that would undoubtedly enrich any environment. I don't hesitate to say that Hardik will be a valuable asset to any organization.
CEO-Infomatic Solutions | Former Associate Professor - Nirma University
Hardik has proven himself to be a thorough researcher and has worked hard extensively to achieve his goals. Totally, Hardik is an extraordinary student and a true credit to any programme he may join.
Manufacturing Manager at Unilever
Hardik is an exemplary candidate who has been a pleasure to work with. Hardik is a hardworking and smart employee. Hardik has proven himself time and again that he is dependable. Hardik can connect easily with shop-floor employees and can communicate with ease to any high or low tier employees proposing any project solutions in accordance with business demands. He has earned the faith, trust and respect of all the shop-floor employees and that of his colleagues. Hardik has been proactive and took the initiative to work on the Auto wrapper reel changeover project.
Get in Touch
If you have scrolled down till here, then probably some part of my profile interests you. Let's connect and take it from there. I am thrilled about our potential collaboration and I look forward to it. Please feel free to reach out to me on any platform that you feel most comfortable with. Thank you!
- Santa Barbara, CA, USA 90007
- (213) 414-8703
- hp.6318@gmail.com
