OdinEYE

Modern software development teams generate large amounts of activity across platforms such as GitHub and Jira, but this information is often fragmented across different tools. Engineering teams may struggle to obtain a unified view of development performance, identify workflow bottlenecks, and understand how development activities affect delivery efficiency.

Traditional metrics also provide limited visibility into potential issues such as pull requests becoming stale or developing a high risk of delayed completion. Teams need a way to transform development activity into actionable engineering insights without relying entirely on manual analysis.

OdinEYE addresses this problem through a full-stack, event-driven engineering analytics platform that integrates GitHub and Jira activity into a unified system.

The platform uses backend microservices to collect, normalize, process, and distribute development events. An independent Python/FastAPI analytics service processes development activity, calculates DORA metrics, and generates Developer Experience insights.

The analytics layer also incorporates machine learning models, including Random Forest, Logistic Regression, and XGBoost, to identify stale and high-risk pull requests from historical development patterns. This enables teams to move beyond descriptive metrics toward proactive identification of potential development workflow issues.

OdinEYE is a full-stack engineering analytics platform built to provide development teams with a unified, data-driven view of their software engineering workflows.

The system integrates activity from GitHub and Jira and processes these events through an event-driven microservices architecture. Incoming activity is normalized into a consistent event representation before being published for downstream services. This allows different components of the platform to independently consume and process development events while maintaining separation of responsibilities.

The backend is implemented using Java and Spring Boot for core platform services, while Python and FastAPI are used for the independent analytics and machine learning service. PostgreSQL is used for persistent data storage, with Docker supporting containerized development and deployment. RabbitMQ is used for asynchronous event communication between services, and AWS is used as part of the deployment infrastructure.

A key component of OdinEYE is its analytics service. Development activity is processed using Python, Pandas, and NumPy to calculate engineering performance metrics, including DORA metrics, and generate Developer Experience insights. These analytics provide teams with a clearer understanding of software delivery performance and development workflows.

OdinEYE further extends traditional engineering analytics with machine learning. Historical development patterns are used to train models such as Random Forest, Logistic Regression, and XGBoost to identify stale and high-risk pull requests. This creates the foundation for proactive engineering insights, allowing teams to identify potential workflow problems earlier.

The project demonstrates the integration of microservices, event-driven systems, software engineering analytics, machine learning, containerization, and cloud deployment into a single production-oriented platform.

Key Areas:

GitHub and Jira activity integration Event-driven microservices architecture Development activity normalization and processing DORA metrics calculation Developer Experience analytics Machine learning-based pull request risk prediction Asynchronous communication using RabbitMQ Containerized services using Docker Cloud deployment using AWS Independent analytics service using Python and FastAPI Future Work

Future improvements include expanding predictive analytics, introducing additional engineering health indicators, improving model accuracy with larger historical datasets, and providing more proactive recommendations based on detected development patterns.