JOSHUA DEAN SAMJAYA

• Energy-saving rApp + traffic-steering rApp → up to 17% gNB energy reduction; PlugFest-validated.
• Time-series predictor (LSTM/XGBoost/CatBoost/RandomForest) with sub-second inference; integrated into rApp policies.
• React + Python API + InfluxDB + Grafana stack; real-time KPIs and admin controls (audit log).
• Multi-vendor E2/xApp interoperability; authored test plans and defect triage to demo readiness.
• Collaborated with a team of 3 to develop and patent an ORAN-compliant energy-saving rApp with traffic-steering rApp in an end-to-end test scenario involving gNBs and Near-RT RIC and Non-RT RIC and SMO to achieve an energy-saving of up to 17%.
• Independently designed and trained AI-based gNB traffic predictor in a 5G environment, integrated within end-to-end 5G environment utilising both commercial and open-source products.
• Independently developed an end-to-end scalable and containerized full stack solution, from front-end GUI with React+JSS and backend and database solution involving SQL and InfluxDB for real-time monitoring.
• Worked with multiple local and international vendor and mobile network providers, driving vendor collaboration and ORAN certification via biannual regional PlugFests to demonstrate vendors and providers’ product maturity and readiness for public use.

● Collaborated with a team of 3 to provide on-the-clock monitoring and issues resolution in Huawei's DWDM and IP equipment, ranging from on-site hardware to cloud-based software and in-house network monitoring system.

● Directly plan, oversee, and handle the national deployment of new Huawei equipment via remote and on-site configuration, including designing and fine-tuning of parameters.

● Independently conducted periodic health checks, preventive maintenance, and software and physical hardware updates on over 1000 Huawei equipment to ensure reliable functionality and prevent service disruptions.

● Executed troubleshooting tasks with 100% efficiency and resolution rate with various clients, and completed over 100 equipment orders and service requests to fulfill promised service level agreements.

• Collaborated with a team of 4 to develop an embedded STM32Cube.AI model to recognize eight different simple hand gestures.
• Enabled efficient prototyping and validation of the STGesture Computer Vision AI model from initial idea to deployment.
• Designed a Python script to automate the processing of collected raw hand gesture data.
• Executed the training, tuning, and testing of the AI model in Python using Tensorflow and Keras to achieve an accuracy of 70% with the model, exceeding the research team’s expectations.