Parcel Tracker

Parcel tracker – a complex software solution for carriers.

Goal

To develop a solution that would:

• Allow carriers to analyze, control, and keep records of parcels.
• Automate delivery tunnels.
• Facilitate opening new destinations.
• Reduce issues during transportation.
• Increase customer base and customer satisfaction.

Result

The software:

• Supports the described above features. 
• Enables users to track parcel location via the web interface.
• Allows users to build workflow reports and metrics based on the collected data and display summarized information in graphs, tables, charts.
• Provides integration opportunities with external platforms (i.e., Splunk, self-developed interfaces, etc.) for further data analysis.​

Technologies

• Splunk
• Hazelcast
• Python, Java, JavaScript
• Spring (Web, Security, Integration), Apache Mina, Apache Camel
• Tomcat
• MongoDB
• Jenkins, Ansible

AI Technologies

• deployment of AI Models using: triton nvidia server and tensorflow-serving​

Goal: object detection and classification on images​

• handling image by multi-model pipelines using python Flask application or DeepStream NVIDIA server​

Goal: adaptation of ai-models results to business requirements​

• implementation of interface to train ai-models using cloud services: Google VertexAI and Valohai​

Goal: creating ai models to handle image by object detection or classification​

• training with local environment: EfficientDet, EfficientNet and Yolov7 models​

Goal: research and providing new ai models​

• ai models optimization using: NVIDIA TensorRT and Intel OpenVino​

Goal: improving execution time and decreasing resources usage​

• implementation of models validator by PASCAL Visual Object Classes requirements​

Goal: regular model accuracy assessment​

• technical support for data analysis using Jupyter, Pandas, Plotly, MathPlotlib​

Goal: data change assessment​

​