Alerting module for a food production line monitooring SAAS

Context

Dataswati develops a Software-as-a-Service designed for the food processing industry. It collects data from various sources such as sensors, settings, ingredients, and recipes and uses advanced analytics to reduce resource consumption while improving product quality.

The Alert Module has been designed as a notification mechanism for the production line, providing critical information regarding impending challenges or those already occurring. Implementing cutting-edge technology, this robust system serves to proactively address concerns before they escalate or rapidly respond to current obstacles. Through its flawless integration within the system, the Alert Module effectively reduces operation interruption, enhancing overall performance and efficiency.

If you wish to learn more about how this SAAS could benefit your organization, please visit the official website, you can ask for a demo on the website or or email the CEO directly at averleyen@dataswati.com to schedule a customized demonstration.

Methodology

Our product offers diverse categories of alerts, including rudimentary ones that monitor preset parameters to ensure stability, along with intricate versions utilizing sophisticated machine learning techniques such as forecasting, classification analyses, supervised/unsupervised learning, etc., maximizing accuracy.

By incorporating an intelligently configured AutoML workflow, we facilitate easy adaptation to match your project requirements. Simply select the classification type and desired features, and our algorithm will generate high-quality models by analyzing combined time series, evaluating them against corresponding validation sets, and selecting the strongest performer for reliable predictions. Periodic reassessments ensure continuous effectiveness, supplemented by detailed reports on alert performance, offering transparency into the system’s behavior. To encourage interactive communication with end-users and further optimize outcomes, our platform enables manual verification or dismissal of triggered alerts. This iterative process fosters cooperation between human intuition and automated decision-making processes, driving improvements through personalized suggestions extracted from refined statistical insights gathered via SHAP value alterations. By balancing user involvement and cutting-edge tech, our system harmoniously blends human expertise with artificial intelligence to achieve successful detection and resolution of potential problems, ultimately leading to improved efficiency across target industries.

Tools

• MongoDB as the main database
• Python : pandas, LightGBM, scikit-learn, shap, optuna, mlflow
• Airflow for scheduling and pipelining
• Docker for packaging
• VSCode for development

My involvment

As the lead for this venture, my responsibilities encompassed every stage, ranging from Interactive Human Machine (IHM) interface conception to management dialogues concerning the commodity and customer interactions. My duties also entailed heading the development staff, formulating the software program, and developing the ML algorithms. Throughout the course of the project, I guided and coordinated efforts to bring forth effective results fulfilling the expectations of both internal entities and external clients.

Results and achievements

This pioneering project marked Dataswati’s initial foray into operationalizing AI in real-time environments, paving the way for future initiatives. It allowed a client to undertake experimental trials on an active production line, tackling recurrent challenges exposed by the custom alerts. As a result, it strengthened our position as a supplier of innovative AI-powered industrial services, bolstered by the combination of advanced technology and exceptional support.