Senior Machine Learning Engineer

• Jan 2023 - Present

Implemented supervised algorithms such as K-means,DBScan clustering for segmentation of patients and location detection . Constructed feature engineering pipelines to extract meaningful features using raw GPS metrics data (Calculate Phone usage ,No of locations visited ,Speed/mean/variance ,Home stay detection). Performed exploratory data analysis activities including visualizing distributions of variables and correlations between them. Researched machine learning algorithms and… Show more Implemented supervised algorithms such as K-means,DBScan clustering for segmentation of patients and location detection . Constructed feature engineering pipelines to extract meaningful features using raw GPS metrics data (Calculate Phone usage ,No of locations visited ,Speed/mean/variance ,Home stay detection). Performed exploratory data analysis activities including visualizing distributions of variables and correlations between them. Researched machine learning algorithms and tools to predict Depression score and determine optimal systems for implementation. Tuned hyperparameters to optimize performance of supervised learning models such as Random Forest, SVM, XGBoost via grid search optimisation. Evaluated model performance metrics such as precision, recall, ROC curves, AUC scores using Patients activities data Show less Implemented supervised algorithms such as K-means,DBScan clustering for segmentation of patients and location detection . Constructed feature engineering pipelines to extract meaningful features using raw GPS metrics data (Calculate Phone usage ,No of locations visited ,Speed/mean/variance ,Home stay detection). Performed exploratory data analysis activities including visualizing distributions of variables and correlations between them. Researched machine learning algorithms and… Show more Implemented supervised algorithms such as K-means,DBScan clustering for segmentation of patients and location detection . Constructed feature engineering pipelines to extract meaningful features using raw GPS metrics data (Calculate Phone usage ,No of locations visited ,Speed/mean/variance ,Home stay detection). Performed exploratory data analysis activities including visualizing distributions of variables and correlations between them. Researched machine learning algorithms and tools to predict Depression score and determine optimal systems for implementation. Tuned hyperparameters to optimize performance of supervised learning models such as Random Forest, SVM, XGBoost via grid search optimisation. Evaluated model performance metrics such as precision, recall, ROC curves, AUC scores using Patients activities data Show less

Data Scientist II (R&D Manufacturing)

• May 2020 - Dec 2022

Cosmetic Audit-AI for wipers- Replace manual black wipe cosmetic audit led by auditor with automated system used by operators. Results were 98% accurate and is deployed in the production. Seagate Enhance Monitoring System(Research) - use of Lidar to provide wider coverage of monitoring of an area using AI/ML/DL and to apply in edge device like Jetson Nano. (3D object detection algorithms) NPS Marketing Analytics (NLP) – developed a machine learning model to find out the sentiment by… Show more Cosmetic Audit-AI for wipers- Replace manual black wipe cosmetic audit led by auditor with automated system used by operators. Results were 98% accurate and is deployed in the production. Seagate Enhance Monitoring System(Research) - use of Lidar to provide wider coverage of monitoring of an area using AI/ML/DL and to apply in edge device like Jetson Nano. (3D object detection algorithms) NPS Marketing Analytics (NLP) – developed a machine learning model to find out the sentiment by topics (e.g., support, products, etc..) from feedback reviews to guide Marketing team to look at the areas where could be improved and suggest insights/actions to improve the customer experience/sentiment . Synergos -(Federated Learning )Worked in collaboration with AI Singapore to create a Global model by connecting machine learning models to the disjointed data required to effectively train them regardless of their locations. Sputter Valve Leak detection Have created the Kibana dashboard which reduce the time taken for diagnosis of sputter leak. Have diagnose missing station in sputter leak analysis ave improved the efficacy of the algorithm CI/CD pipeline to deploy services on Kubernetes Implemented Direct Acyclic Graphs (DAGs) in Apache Airflow to schedule and monitor workflows of leak alert system Mentored junior team members on best practices for conducting effective data analysis in the context of healthcare operations. Developed training materials that help users(NON-IT Engineers) understand the basics of machine learning concepts. Mentored interns on how to interpret results from predictive/data analytics projects Show less Cosmetic Audit-AI for wipers- Replace manual black wipe cosmetic audit led by auditor with automated system used by operators. Results were 98% accurate and is deployed in the production. Seagate Enhance Monitoring System(Research) - use of Lidar to provide wider coverage of monitoring of an area using AI/ML/DL and to apply in edge device like Jetson Nano. (3D object detection algorithms) NPS Marketing Analytics (NLP) – developed a machine learning model to find out the sentiment by… Show more Cosmetic Audit-AI for wipers- Replace manual black wipe cosmetic audit led by auditor with automated system used by operators. Results were 98% accurate and is deployed in the production. Seagate Enhance Monitoring System(Research) - use of Lidar to provide wider coverage of monitoring of an area using AI/ML/DL and to apply in edge device like Jetson Nano. (3D object detection algorithms) NPS Marketing Analytics (NLP) – developed a machine learning model to find out the sentiment by topics (e.g., support, products, etc..) from feedback reviews to guide Marketing team to look at the areas where could be improved and suggest insights/actions to improve the customer experience/sentiment . Synergos -(Federated Learning )Worked in collaboration with AI Singapore to create a Global model by connecting machine learning models to the disjointed data required to effectively train them regardless of their locations. Sputter Valve Leak detection Have created the Kibana dashboard which reduce the time taken for diagnosis of sputter leak. Have diagnose missing station in sputter leak analysis ave improved the efficacy of the algorithm CI/CD pipeline to deploy services on Kubernetes Implemented Direct Acyclic Graphs (DAGs) in Apache Airflow to schedule and monitor workflows of leak alert system Mentored junior team members on best practices for conducting effective data analysis in the context of healthcare operations. Developed training materials that help users(NON-IT Engineers) understand the basics of machine learning concepts. Mentored interns on how to interpret results from predictive/data analytics projects Show less

Data Scientist(Advance Analytics)

• Jun 2019 - May 2020

Finance Forecasting -Client Microsoft Operations Lmt (One Finance): Collaborated with stakeholders to evaluate business problems and develop appropriate AI and ML solutions. Applied statistical methods like hypothesis testing, ANOVA tests, for validating research findings Tested different feature engineering strategies such as dimensionality reduction and feature selection. Evaluated accuracy of machine learning models using metrics like precision and recall scores or F1… Show more Finance Forecasting -Client Microsoft Operations Lmt (One Finance): Collaborated with stakeholders to evaluate business problems and develop appropriate AI and ML solutions. Applied statistical methods like hypothesis testing, ANOVA tests, for validating research findings Tested different feature engineering strategies such as dimensionality reduction and feature selection. Evaluated accuracy of machine learning models using metrics like precision and recall scores or F1 scores. Deployed web applications with Flask and Django to provide interactive access to models and visualisations. Conducted natural language processing tasks such as sentiment analysis and text classification. Built dashboards with Tableau to visualize complex datasets and report results of analyses' Designed experiments to test hypotheses and evaluate model performance metrics. Show less Finance Forecasting -Client Microsoft Operations Lmt (One Finance): Collaborated with stakeholders to evaluate business problems and develop appropriate AI and ML solutions. Applied statistical methods like hypothesis testing, ANOVA tests, for validating research findings Tested different feature engineering strategies such as dimensionality reduction and feature selection. Evaluated accuracy of machine learning models using metrics like precision and recall scores or F1… Show more Finance Forecasting -Client Microsoft Operations Lmt (One Finance): Collaborated with stakeholders to evaluate business problems and develop appropriate AI and ML solutions. Applied statistical methods like hypothesis testing, ANOVA tests, for validating research findings Tested different feature engineering strategies such as dimensionality reduction and feature selection. Evaluated accuracy of machine learning models using metrics like precision and recall scores or F1 scores. Deployed web applications with Flask and Django to provide interactive access to models and visualisations. Conducted natural language processing tasks such as sentiment analysis and text classification. Built dashboards with Tableau to visualize complex datasets and report results of analyses' Designed experiments to test hypotheses and evaluate model performance metrics. Show less

Research Intern Deep Learning

• Aug 2018 - Mar 2019

Fault detection and Classification in Sensors SemiConductor Industry To improve accuracy level for detecting faulty sensors and to reduce human effort using Computer vision and deep learning Image Enhancement Cropping each image to only keep the sensor without Background Implemented various custom loss functions like Weighted Loss to improve U-net segmentation and have used Multi Class Segmentation (Unet 2D ) Experimented with various Convolution Neural Networks and Optimised the… Show more Fault detection and Classification in Sensors SemiConductor Industry To improve accuracy level for detecting faulty sensors and to reduce human effort using Computer vision and deep learning Image Enhancement Cropping each image to only keep the sensor without Background Implemented various custom loss functions like Weighted Loss to improve U-net segmentation and have used Multi Class Segmentation (Unet 2D ) Experimented with various Convolution Neural Networks and Optimised the results by Tuning Parameters and by Cross Validation Segmentation and Classification based on the shape, size and the position of the defect. Deep Learning For Medical Image Analysis Blood Vessel Segmentation of Heart: The Blood vessel (Aorta ) has to be extracted from all other parts in CT scan data using various 3D Convolution Neural Network Architecture which would be used for further clinical reference. Metal Artefacts Reduction in CT reconstruction using deep learning-Thesis : we develop a convolution neural network (CNN) based Metal Artifact Reduction framework to suppress artefacts caused due to Metal Implants in Xray CT images Show less Fault detection and Classification in Sensors SemiConductor Industry To improve accuracy level for detecting faulty sensors and to reduce human effort using Computer vision and deep learning Image Enhancement Cropping each image to only keep the sensor without Background Implemented various custom loss functions like Weighted Loss to improve U-net segmentation and have used Multi Class Segmentation (Unet 2D ) Experimented with various Convolution Neural Networks and Optimised the… Show more Fault detection and Classification in Sensors SemiConductor Industry To improve accuracy level for detecting faulty sensors and to reduce human effort using Computer vision and deep learning Image Enhancement Cropping each image to only keep the sensor without Background Implemented various custom loss functions like Weighted Loss to improve U-net segmentation and have used Multi Class Segmentation (Unet 2D ) Experimented with various Convolution Neural Networks and Optimised the results by Tuning Parameters and by Cross Validation Segmentation and Classification based on the shape, size and the position of the defect. Deep Learning For Medical Image Analysis Blood Vessel Segmentation of Heart: The Blood vessel (Aorta ) has to be extracted from all other parts in CT scan data using various 3D Convolution Neural Network Architecture which would be used for further clinical reference. Metal Artefacts Reduction in CT reconstruction using deep learning-Thesis : we develop a convolution neural network (CNN) based Metal Artifact Reduction framework to suppress artefacts caused due to Metal Implants in Xray CT images Show less

Software Engineer | Data analyst

• May 2016 - Dec 2017

SAP BO Developer and Business Analyst( Anthem Wellpoint(Contact Center Reporting)): 1.Engage with the client to gather functional and non functional requirements to enhance the design of BI deliverable(Reports,dashboards) using SAP Business objects 2.Experience in creating dashboard using Excel,Tableau to analyse Key metrics and maximize the revenue growth.

Associate Software Engineer

• May 2016 - Nov 2017