skills
experience
Medical Imaging Data Engineer co-op
Philips North America
Orchestrated the preparation and preprocessing of ultrasound image data for Project iMAMA, a Gates foundation-funded machine learning initiative focused on obstetric ultrasound diagnostics aimed at improving patient care
Deployed YOLO model on unstructured dataset of ultrasound image data for detection of fetal presence for prediction of features like gestational age and number of gestations, contributing to 45% accuracy in object detection
Used data exploration and transformation techniques to transform 3TB+ of unprocessed ultrasound image data into structured NumPy arrays, enabling estimation of ultrasound features using predictive analytics reducing manual labor by 10% each month
Migrated 15TB+ data warehouse including stored procedures, triggers and ETLs to MD.ai (a medical AI cloud healthcare API) leveraging AWS S3 and Glue resulting in yearly cost savings of $25k
Automated a data pipeline using Python and Airflow to catalog ultrasound information of 100k+ patients for annotation extraction purposes, saving 15 hours of manual labor each week
Partnered with doctors on clinical team to develop clinical review platform on MD.ai for annotating the migrated ultrasound images to be further extracted periodically (every 48 hours) by means of the Airflow DAG
Collaborated with 3 cross-functional teams to integrate data from 9 disparate data sources using Python, S3 and Redshift to create BI-ready data views for analysis with AWS Quicksight dashboard to derive clear, compelling actionable insights
January 2023 - August 2023
Application Development Analyst
Accenture
Utilized python libraries to outline anomalous securities by applying statistical approach (Z-score and Standard Deviation); received 84% approval from Compliance team
Trained model to apply anomaly detection algorithm to assign anomaly scores and filter out anomalous securities with accuracy of 78% contributing to development of marketing mix models and delivering 10% boost on ROI
Performed scripting using python to fetch attachments/data reports from Outlook and upload it into Snowflake
Developed interactive dashboards for deeper trend analysis and reported anomalies to Compliance team on weekly basis for inspection
Programmed to schedule a Python job to upload data in Snowflake and send automated email alerts everyday
June 2021 - December 2021
projects
Developed Large Language Model (LLM) using Optical Character Recognition (OCR) technology on various use cases – images containing text with different font, sentiment analysis, hand written notes to extract text from images. Leveraged transformers to utilize pre-trained LLM models like BART for various Natural Language Processing (NLP) tasks, enhancing accuracy of the extracted text output
#Python
#Pandas
#Numpy
#OpenCV
#Tesseract
#Transformers
#Neural Networks
Conducted comprehensive pre-processing and imputation on financial dataset to determine if client has defaulted in monthly credit card payments, interpreting findings through visualization and advanced statistical techniques. Developed and evaluated various machine learning models including logistic regression, SVM and Random Forest classifier, identifying SVM as the optimal model with 86% classification accuracy, improving efficiency by 17% compared to baseline model.
#Python
#Pandas
#Numpy
#Scikitlearn
#Tensorflow
#Seaborn
#Matplotlib
Implemented a flight classification model based on the flight delay factor employing a dataset encompassing flight data recorded in January 2019 in the United States consisting of 583,986 records and 22 binary categorical data attributes. Performed intricate cleaning and preprocessing operations and performed exploratory data analysis (EDA) to identify trends and make interpretations and optimizations on a logistic regression base model with classification accuracy of 85%. Employed 5+ advanced machine learning algorithms with highest classification accuracy of 92% on Multi Layer Pereptron (MLP) Neural Network, improving data efficiency by 7%
#Python
#Pandas
#Numpy
#Scikitlearn
#Seaborn
#Matplotlib
