Features

● Capture data from sensors in mining machines spread across multiple sites

● Sources: Sensors placed near components within the machine

● Each sensor type has alarm and shutdown levels

● Plot of average sensor values at different levels with drill-down

● Plot of min and max in sensor values in the past month

● Machine Learning: Predictive analytics on machine failure and component
maintenance/replacement based on historical data

Big Data Analytics Stack

● Airflow, Argo

● Sqoop, Kafka, Spark
● HDFS, Minio, SeaweedFS
● Hive
● Mongodb, Cassandra, Aerospike, ElasticSearch, PrestoDB, Druid
● Kubernetes, Docker
● Superset, Metabase
● PowerBI, Tableau

Machine Learning

● Descriptive and Predictive analysis of structured data - Risk analysis, Fraud detection,
Security incident detection
● Computer vision - Object detection based signature detection in scanned documents.
Image-based document classification and extraction, License-plate detection
● NLP - Named entity recognition, Sequence to Sequence modelling, Sentiment analysis,
Text-based classification

Technology Stack

● TensorFlow, KubeFlow, PyTorch, AutoML, MLlib, Dask, Cloud ML, BERT
● Amazon, Google, Microsoft, Linode, Paperspace