CV

Education

Software Engineer II - Machine Learning
Philips
April 2021 - July 2023

Deployed Machine Learning models for automated customer complaints classification using R and NLP, to achieve 82% automation coverage and reduce manual classification time from 2 days to 10 minutes.
Led a team of 5 engineers to develop and deploy MR medical imaging pipelines across 50+ hospitals, using Computer Vision to enhance image clarity and resolution, resulting in 20% faster and more accurate diagnoses.
Engineered predictive models to accurately detect failures in critical components like magnet coils and X-ray tubes, reducing unplanned downtime by 50 incidents annually and saving 500 crucial person-hours.
Leveraged time series forecasting to drive utilization insights across the MR installed base, improving resource allocation by 25% and operational efficiency by 15%, resulting in $1 million annual cost savings.
Developed the translator microservice using Flask to achieve translation retrieval times of 2 milliseconds using Redis caching, enabling real-time translation of global service calls and automating NLP-driven issue classification.

Software Engineer I
Philips
July, 2019 - March, 2021

Mastered functional programming language Erlang to lead the resolution of critical bugs in the Data Manager module, reducing system downtime by 40% and enhancing reliability for 30+ healthcare facilities.
Streamlined radiology operations by integrating HL7, DICOM and scheduling data into BI tools for actionable insights.
Implemented the bulk import feature for the service tools application of the PerformanceBridge platform, enhancing data processing efficiency by 21% and reducing manual entry errors by 78%, impacting the workflow of 300+ users.

Software Engineer Intern
PayPal
July, 2018 - Dec, 2018

Developed an auto-remediation framework, Optimus, to resolve data operations failures using ML algorithms, reducing downtime by 35% and saving $200K annually in operational costs for PayPal.
Created a full-stack web application using Spring framework to monitor and resolve failures in Hadoop, RabbitMQ and Spark systems, to improve the reliability of data processing workflows for over 500,000 transactions daily.

Advanced Camouflaged Object Detection in Limited Data Setting
Python | Computer Vision | April - May 2024

Modified the SINet using ResNet-18 to reduce computational load while retaining 90.76% of detection accuracy.
Improved pixel accuracy of segmentation masks from 69.19% to 79.22% on COD10K by introducing balanced loss.
Implemented data augmentation techniques using style transfer & stable diffusion API to simulate natural camouflage.
Tested with 3,000 original, style-transferred, and synthetic images to achieve 11.36% increase in detection precision.

Prompt Score
Python, PyTorch, Huggingface, Llama | Natural Language Processing | Mar - May 2024

Developed a prompt scoring system to evaluate the specificity of prompts used with Large Language Models (LLMs).
Fine-tuned the Llama model with QLoRA to achieve a 20% improvement in performance on specificity scoring.
Evaluated 3 LLMs (Alpaca, Claude, and Gemma) to find that Claude was 15% better in coherence and constraints.
Created and annotated a dataset of 800 prompts, improving prompt scoring model accuracy by 10%.

Programming Languages: Python, Java, C, C++, HTML/CSS (Frontend), JavaScript(Backend), SQL
Libraries: PyTorch, TensorFlow, Scikit-learn, Pandas, Keras, OpenCV
Technologies/Frameworks: Flask, Spring, React, Node.js, ExpressJS (MERN)
DevOps: Git, GitHub (CI/CD), Talend, Jenkins
Databases: MongoDB, Vertica, MySQL, PostgreSQL
Other Skills/Tools: Redis, Kafka, Docker, Kubernetes, Hive, Spark