Data-Driven Insights

Transforming manufacturing with advanced models for predictive maintenance and anomaly detection solutions.

Data Platform

Multi-source data platform construction for predictive maintenance solutions.

The image depicts a close-up view of industrial machinery, highlighting a section of an engine with various components, including bolts, pipes, and a small glass container with a yellowish liquid. The machinery appears to be metallic with a weathered surface, suggesting regular use.
The image depicts a close-up view of industrial machinery, highlighting a section of an engine with various components, including bolts, pipes, and a small glass container with a yellowish liquid. The machinery appears to be metallic with a weathered surface, suggesting regular use.
Model Training

Training LSTM, TFT, and autoencoder models for predictive maintenance insights.

An industrial setting inside a factory with multiple large mechanical components arranged in a row. A complex network of beams, pipes, and metal structures is visible above, indicating a manufacturing or assembly environment. The lighting is dim and utilitarian, focusing on functionality rather than aesthetics.
An industrial setting inside a factory with multiple large mechanical components arranged in a row. A complex network of beams, pipes, and metal structures is visible above, indicating a manufacturing or assembly environment. The lighting is dim and utilitarian, focusing on functionality rather than aesthetics.
Advanced Integration

Implementing retrieval-augmented techniques for enhanced maintenance decision-making processes.

“Survey of Deep Learning for Industrial Machine Fault Prediction” (2021, lead author)

Reviewed time‐series models, autoencoders, and GANs in fault detection, informing our architecture choices.

“Self‐Supervised Pretraining for Time‐Series Forecasting” (2022, co‐author, NeurIPS Workshop)

Introduced TSTPretrain, pretraining transformers on unlabeled data to significantly boost downstream PdM accuracy.

Data

Integrating sensors and models for predictive maintenance.

© 2025. All rights reserved.