Jaskirat Bhatia

I am an accomplished and highly skilled Machine Learning Engineer with a strong academic background. My Master's degree in Data Science and AI from the University of Waterloo showcases my dedication to gaining comprehensive knowledge in the field. My professional experience at Extend AI demonstrates my expertise in computer vision and deep learning. My ability to lead and communicate effectively was evident in my role at iMerit, where I managed a team of 5 annotators for a complex data annotation project. Additionally, I showcased my problem-solving abilities and technical proficiency by designing the auto-ML solution "Spacetime University" which streamlined training processes and reduced training time. My research experience, such as predicting molecule binding and discovering drugs, highlights my ability to apply machine learning techniques to real-world problems. My diverse skill set, which includes programming languages like Python and R, along with libraries/tools like PyTorch, OpenCV, and MongoDB, further demonstrates my versatility. Overall, my achievements and experiences reflect my passion for machine learning, and my ability to deliver impactful solutions to complex challenges.

Throughout my professional journey, I have gained valuable experience as a Machine Learning Engineer, specializing in Computer Vision and training Deep Learning models at Extend AI. This role has allowed me to leverage my solid academic background, including a Master's degree in Data Science and AI from the prestigious University of Waterloo. One of the notable projects I collaborated on was the Stone AI project, where I developed a machine learning stack and trained deep learning models using advanced hardware infrastructure. Through this project, we achieved an impressive 80% accuracy in semantically segmenting anomalies in stone slab images. Additionally, I contributed to improving Extend AI's ML solutions by devising and executing a two-step generative style transfer model. This model helped generate data in a known distribution and perform style transfer to mimic data in a missing distribution, thereby enhancing the quality of our ML solutions. I also played a key role in designing "Spacetime University," an auto-ML research and training solution that significantly reduced training time by 40%. At iMerit, I spearheaded a data annotation project involving the annotation of 46 types of stone slabs with 23 different types of anomalies, managing a team of five annotators. By analyzing a dataset of 30,000 high-resolution images, I estimated the required amount of data annotations using statistical inferences like KL Divergence. Furthermore, I proposed and executed a data annotation strategy that delivered over 10,000 image annotations, resulting in a 20% increase in the accuracy of our ML models. Prior to Extend AI, I served as a Research Assistant at the University of Waterloo, where I predicted molecule binding to the Covid-19 binding site and discovered 242 drugs, including Remdesivir and Hydroxychloroquine. During my internship at InterGlobe Technologies, I successfully improved an email classifier's accuracy from 74% to 87% using a Naive Bayes model based on bag-of-words text modelling. Additionally, I implemented a deep learning model on Twitter data with an 85% accuracy in predicting the sender's mood.

I have studied for a total of 6 years after high school, doing a Bachelor's in Technology(Computer Science) from SRM University and a Master's in Data Science and AI from the University of Waterloo. During my study at Waterloo, I took on 9 courses: Classification, Data Visualization, Data Exploration, AI, Statistics, Big Data, Optimization Techniques, Deep Learning and Causal Inference. Apart from the above, I completed various projects including: Street Sign Classifier, Happiness Index, Study on how Context Matters, etc.