G Mothy | AIcrew Stories
🎉 Welcome back to AIcrew Stories!
Their stories are sure to inspire beginners in Machine Learning. Read on to get some guidance, learn new tools and tricks. Maybe someday you will have the story of your journey up here!
🙋🏼♂️ Let’s meet G Mothy
G Mothy hails from Hyderabad, India. He has completed his undergraduate degree from the Army Institute of Science and Technology and currently works as a Machine Learning Engineer at Quantiphi. He works on solving real-life problems in industries using Computer Vision and Machine Learning.
AI Blitz 7 was Mothy's debut challenge on AIcrowds' platform.
🚗 Mothy's Machine Learning Journey
Mothy’s machine learning journey started when he took up an internship at IIT Madras in his sophomore year. During this period, he interacted with like-minded Ph.D. students who worked on solving data engineering problems. He also got first-hand experience working on GPU and hardware integration.
Since then, with the guidance of his fellow seniors and the support of his college mates, Mothy built upon his existing knowledge in the field of Artificial Intelligence and Machine Learning. Through ML competitions, Mothy eventually got the hang of Hackathons and Machine Learning challenges. After all the hard work, seeing his name on the leaderboard at the end of the challenge was an exhilarating experience.
💪🏼 Getting Started with the Challenge
Looking for ML challenges to seize, Mothy came across AI Blitz. “I found the format of AI Blitz to be quite intriguing as these 5 AI puzzles covered all the major domains of Computer Vision”.
With this challenge covering all the major Computer Vision tasks like detection, classification, and image correction, Mothy considered it a comprehensive revision of these concepts.
🧭 His Approach
Tackling the prompts one by one, Mothy first decided to "Blitz" the Rover Classification challenge. For this he started with the baseline model provided by AIcrowd, Mothy was able to make edits and omissions in the model pipeline for a better solution. One such amendment was switching the existing AlexNet architecture and utilizing the skip connections of the ResNet50, through which Mothy was able to achieve the best score possible. Following the workflow of the previous task, Mothy implemented the same approach of iterative development of the baseline model with state-of-the-art architectures.
This approach took Mothy around 16 hours. His hard work and commitment ultimately helped him conquer the challenge and emerge as the winner.
Apart from learning more about real-life applications of Machine Learning, Mothy is interested in business. Like many of us, in his leisure hours, he likes to read books and watch movies.