Your strategic partner for end-to-end AI solutions

Machine learning-enabled anomaly detection for the Canadian Space Agency (CSA)

Calian completed a project for the CSA that uses AI to automate anomaly detection capabilities for satellite telemetry data. The solution used operational satellite telemetry data to train an ensemble of deep neural networks (LSTMs) to learn to detect anomalies. It’s estimated to save about 65 per cent of an operator’s time spent on FDIR process.  

AI-driven solution for NATO—generating realistic data for exercises

Calian partnered with NATO’s Joint Assessments Division to deliver a cutting-edge AI engine that automatically generates the realistic data and insight reports analysts need during training exercises. Leveraging our deep defence expertise and advanced AI pedigree, we built dynamic, context-aware pipelines that fuse political, military, economic, social, information and infrastructure (PMESII) factors with diplomatic, informational, military, economic, financial, intelligence and legal (DIMEFILII) dimensions. The result is a state-of-the-art system that models complex operational environments in real time, giving NATO planners richer scenarios, faster decision support, and more effective training outcomes. 

AI-enabled process to develop content for large, complex military exercises 

Calian has developed AI assistant functionality in MaestroEDE to automate the development of exercise content for any exercise and integrates with the Virtual Command and Control (VCCI) toolset. The AI assistant develops expert-guided draft content to meet the objectives of the exercise, which human operators then verify to ensure it meets training needs and can be used in an exercise. This AI-enabled process reduces the time required to develop content for large, complex military exercises. Accelerating content generation is a game-changer for developing large, complex military training exercises and brings a level of agility and flexibility for development that human operators alone can’t match.  

Proof of concept: analyzing shift logs for predictive analysis  

A multi-departmental Calian team, comprising advanced technologies, emerging technologies and nuclear, built a proof of concept to analyze shift log data to predict and analyze failures or maintenance issues. This required digesting multi-modal irregular data from various data inputs. The team used large language models (LLMs), vector storage and a multi-agent approach to provide actionable insights to operators. 

Applying machine learning to enhance the effectiveness of information from sensors in grain bins

Calian developed a noise reduction application for a Calian agriculture product that monitors grain storage bins. Bin-Sense® uses sensors to detect the level of coverage, temperature and humidity in the bins and provides the information to farmers so that they can monitor their grain. 

This reduces the effort of having a single operator look at all the readings to find false positives. The model generates a coverage prediction that is more accurate than individual readings. Users have increased confidence in the readings that they are receiving, improving efficiency and reducing waste. It also reduces the number of alerts going out to the farmers, who have more pressing tasks than sifting through alerts that may be false.

Deep insight for the Toronto property rental market

Calian built a tool that automates the process of collecting competitive rental property information and uses machine learning for demand forecasting, competitive analysis and price prediction.