IoT
<New Co.>
New Co. provides end-to-end structural health monitoring soution for the global problem of aging infrastructure.
New Co. is an Industrial Internet of Things (IIoT) technology, using tiny fiber optic sensors attached to the bridge to accurately measure and estimate structural strain, thermal response, bending, loads, vibration, and corrosion, which are all measures of structural health.
DOT Structural Health Monitoring
Support cloud-based data retrieval tools, data preprocessing, data integrity pipelines, remote and real-time monitoring algorithms (AWS/SQL/Python)
Develop clustering, anomaly detection and data integrity tools for multiple assets (100-200 sensors per site) using spectral/wavelength data
Design and support deployment of fiber optic sensor installation (~100-200 sensors per bridge)
Engage with stakeholders and collaborate with New Co.'s teams across the globe (hardware/software/operation/product teams)
Support new hiring and data science team scaling by conducting interviews, training, onboarding and mentorship
West Gate Tunnel Project
Delivered a real-time monitoring system for three bridges located on Hume Hwy Victoria during superload moves (200-250 ton trucks)
Developed a cloud-based package for the operation team to analyze data, detect and classify superload per permit conditions, generate automated reports for each superload and flag near real-time alerts
Supported the installation and deployment of ~300 FO sensors and provided monitoring system certification, including data check pipelines/delivering monitoring
Conducted software training and knowledge transfer to the operation team - the product was delivered successfully on September 2021 and loads started operating in October 2021
<Strados>
Strados is an Internet of Medical Things (IoMT) technology, and is the world’s first FDA-cleared wearable lung device for respiratory health monitoring. Strados provides a smart sensor platform to enable early, remote detection of key changes in lung acoustics and ventilation patterns.
Lung Monitoring
Analyzed the Strados device data (microphone, accelerometer and gyroscope) using time/frequency domain analysis (spectrogram and wavelet) for lung health monitoring (cough, wheezing, etc.) and activity detection (not in use, walk, etc.)
Support the development of in-house tools for event annotations, labeling and training
<Siemens>
R&D work at the Autonomous Systems and Control Group
Fault Detection & Localization in Process Industries
Process plant behavior changes with time due to various factors including aging of equipment, malfunctioning in sensors, changed raw material quality
In this project, I worked as the primary researcher to develop hybrid fault detection and localization algorithms for a smart automation plant by modeling hundreds of time-series and contextual signals such as flow, temperature, pressure, level, position
Modeled different processes within the plant’s operation using multi-input-multi-output autoregression models
Improved the defect localization accuracy up to 95% by developing ML algorithms for anomaly detection and fault localization models that combine time-series data and structural information of the plant
Developed a real-time demonstrator and validated it on real data provided by the customer
The outcome of this project was selected for presentation at the AIChE 2020 Conference