Global Catalog - Ingestion and Mapping

Launched a new wait-time model serving traffic in ERDs :

• Wrote batch, streaming and on-demand feature-views that define transformations for feature-computations (Tecton.ai).

• Wrote a Tecton feature-service that exposes computed features at inference time via an API call.

• Unit-tested the feature transformations, and regression tested the feature-service.

• Used Mode + SQL to validate feature computations.

• Trained and evaluated a wait-time model, and exposed it via a Sagemaker endpoint.

• Generated additional Mode reports to identify black-out periods for patient visits.

SWE - Data Science :

• Built-out structured logging for the Data Platform to trace each Sagemaker inference request. Configured a forwarder lambda to send CloudWatch logs to a Datadog Dashboard.

• Ported the platform from using prebuilt scikit-learn Sagemaker containers to Sagemaker Training and Inference Toolkits.

• Wrote unit tests and integration tests for corresponding code changes.

Software Tools COE - ML/AI Working Group

Lead Developer for project Q-Visor - a test case prioritisation tool deployed on Firmware programs across Enterprise (ESS) and Client Storage (CSS) business units.

Responsibilities Include -

• Writing automated ETL to aggregate data from multiple Test Automation Systems, Bitbucket/Gerrit, JIRA and Firmware release jobs.

• Clean, process and persist data.

• Interface with Firmware and Validation teams to understand Firmware-Test interaction.

• Train, optimise and evaluate explainable models for test-failure prediction.

• RESTful integration with Test Automation Systems.