About

Javin Bachani is a Data Science Research Intern at the Massachusetts Institute of Technology (MIT) with a keen interest in applying machine learning algorithms to traditional fields in biology and designing useful software. He is currently pursuing a Bachelor of Engineering in Electronics Engineering and a Master of Science in Biology/Biological Sciences at the Birla Institute of Technology and Science, Pilani. Javin has gained experience in various industries, including his role as a Blockchain Trainee at DRDO, where he set up a permissioned Hyperledger blockchain for multi-organizational secure channel communication and data transfer. He also deployed the network to computers on a Local Area Network using Kubernetes for docker container management and integrated TLS encryption with endorsement policies for secure communication. As a Machine Learning Intern at Plastic Water Labs, Javin created a dataset of 1000 images of normal and defective water bottles using Kaggle and image augmentation techniques. He fine-tuned a transfer learning Mask-RCNN model for object detection and pixellib for image segmentation, created a CNN to classify the segmented water bottle image as normal/defective (as well as type of defect, i.e., scratch or dent) with an 87.7% accuracy. A second CNN was created to detect the material of the bottle (plastic or glass) with a 72% accuracy. The different model weights were saved as an .h5 file using pickle and integrated into a single deep-learning model. Finally, he created a web app for cloud deployment of the model using Flask. Javin's current research at MIT involves investigating the effects of vasopressors on 40,000 patients from the MIMIC-IV database on GCP Bigquery under the supervision of Dr. Celi, Laboratory for Computational Physiology. He evaluates outcomes using Targeted Maximum Likelihood Estimation for causal inference and conceived propensity scores in R using XGBoost and Multivariate regression with 98% accuracy. Javin is a Research Scientist, Data Scientist, and has expertise in Blockchain, XGBoost, ML, GCP, Kubernetes, Docker, CNN, Object detection, Object Detection, Image Segmentation, Flask, and Web. He is always open to opportunities and collaboration and loves conversing with different people, except those from 4chan. Outside of work, Javin is a long-time fan of Real Madrid and makes a killer libero in any volleyball team.