Kartik Pruthi
As a passionate and dedicated Data Science Engineer Intern, I thrive at the intersection of technology and innovation. My primary expertise lies in model compression, data visualization, and model training, particularly focusing on vision-related tasks.
Github
Kartik Pruthi
Kartik Pruthi
About Me
As a passionate and dedicated Data Science Engineer Intern, I thrive at the intersection of technology and innovation. My primary expertise lies in model compression, data visualization, and model training, particularly focusing on vision-related tasks. Through my work, I have gained substantial experience in converting TensorFlow models to TFLite for edge deployment, showcasing my ability to bring cutting-edge machine learning models to resource-constrained environments.
My academic journey at Thapar Institute of Engineering and Technology has provided me with a robust foundation in computer technology, complemented by hands-on experience during my internship at Little Place Labs. Here, I have delved into advanced techniques such as Neural Architecture Search and Low-rank Tensor Decomposition, further honing my skills in optimizing machine learning models.
One of my key projects involves developing innovative solutions for detecting solar panels in drone imagery. This project highlights my problem-solving capabilities and my proficiency in handling complex datasets and training custom models like YOLOv5 and UNet for object detection and segmentation tasks. I take pride in my ability to collaborate effectively with cross-functional teams, leveraging our collective analytical skills to drive successful project outcomes.
My technical skills encompass a broad spectrum, including programming languages such as Python, R, and C++, and frameworks like TensorFlow, Sklearn, and OpenCV. I am adept at working with various platforms, including web, Windows, Linux, and cloud services like AWS. These skills enable me to approach challenges with versatility and adaptability, ensuring that I can contribute to diverse technological environments.
Beyond my technical acumen, I excel in communication, teamwork, and creative problem-solving. I am driven by a passion for contributing to innovative and forward-thinking projects, constantly seeking new challenges that allow me to grow and expand my expertise.
Resume
Career Goals
Short-term Goal: Gain as much knowledge as possible during my master’s studies and secure a job that challenges my problem-solving skills.
Long-term Goal: Provide companies with valuable insights gained during my work years and eventually start a business.
Work Experience
Little Place Labs:
Data Science Engineer Intern | Jan 2024 – Ongoing | New Delhi, India
- Worked on model compression for edge deployment, focusing on TFLite conversions.
- Documented research papers and their abstracts on model compression techniques, most notably Neural Architecture Search and Low-rank Tensor Decomposition.
- Selected and remotely set up an FPGA MPSoC with Pynq Linux image and explored design suits for the same.
- Annotated and trained models to detect solar panels in drone imagery, working on solving the challenge of detecting small-scale objects.
Microsoft Learn Student Chapter:
Executive Member | October 2020 – February 2022 | College Society
- Worked on multiple front-end projects as a team.
- Designed UIs for new websites and posters on various occasions.
Projects
Solar Panel Detection using Drone Imagery:
- Using 30cm resolution images of cityscapes, annotated imagery for solar panels and trained a YOLOv5 custom model for the task, along with a custom-built UNet for segmentation
Wildfire Detection using Satellite Imagery:
- Using Sentinel-2 MSI images of wildfires and forest imagery as training data, trained a U-Net and CNN to classify fire or no-fire as a binary classification.
Emotion Detection from Speech (Signal Processing):
- Using speech data from RAVDESS and SAVEE dataset, trained a Sklearn ensemble to classify a speech/audio clip into four primary emotions.
Detecting Airplanes in Images (Computer Vision):
- Using a custom-annotated small dataset, trained a YOLOv8 model to segment airplanes in fed images.
Gold Price Prediction (Machine Learning):
- Using features such as date and exchange rates, trained a Random Forest ensemble to predict gold prices using tentative features on a tentative date.
Certifications
- Complete Ethical Hacking Bootcamp – Udemy (July ’24)
- Data Analysis for Entry Level Job (9 courses in 1) – Udemy (July ’24)
- Introduction to On-Device AI – Deeplearning.AI (June ’24)
- Machine Learning Specialization – Coursera x Stanford University (December ’22)
- Supervised Machine Learning: Regression and Classification – Coursera x Stanford University (December ’22)
- Advanced Learning Algorithms – Coursera x Stanford University (December ’22)
- Unsupervised Learning, Recommender, Reinforcement Learning – Coursera x Stanford University (December ’22)