Developed a healthcare platform that empowers healthcare researchers with scientifically rigorous, fast and compliant analytics.

* Implemented a scheduling component of a workflow orchestrator (Conductor) that spans across microservices using Spring Boot and PostgreSQL. Built CI/CD pipelines for deployment and dashboards for monitoring.

Pull request is open-sourced: https://github.com/jas34/scheduledwf/pulls

* Built event handlers of the workflow orchestrator for loose coupling between workflows and support for producing and consuming events from Azure Queue Storage in Spring Boot.

* Built the patient de-identification module that transformed patient information in Java. Supported checkpoint feature for failure recovery.

Developed a text analytical engine for customer experience management. Demonstrated continued success during the creation of new analytical and predictive services designed to discover customers sentiment from contact centers to social media online reviews.

* Optimized ElasticSearch queries. Broke down big slow queries into small batches. Stored results in Spring Boot Hazelcast cache per search term. Supported request options (populated percent, distinct count etc.), and only returned data specified in the options.

* Integrated Graphite and Prometheus with applications for monitoring purposes.

Deployed to IBM Cloud Kubernetes Service (IKS) and Amazon Elastic Kubernetes Service (EKS).

* Collaborated with the data science team to build predictive models that predict what makes

a customer promote a product, using h2o machine learning algorithms, such as Generalized Linear Model (GLM) and Gradient Boosting Machine (GBM). Collected algorithm performance statistics from logs for further improvement.

* Designed and implemented ElasticSearch health check. Checked ElasticSearch health before executing queries at index level using cluster health API. Blocked query if index health was poor, logged error, wrote cache.

* Implemented an API that evaluates search terms’ change over time. This new feature enables customers to know if the average for each period is statistically significantly different.

1) Utilized automation platform Chef’s command-line tool knife to retrieve server data, such as node name, server status, run id etc. Scheduled task periodically and wrote them to log files.

2) Post JSON data to Elasticsearch cluster using Chef REST API. Visualized data in Kibana by creating index pattern based on server start time. Dashboard included showing counts of node status: success or failure over time.

1) New features enabled users to know who last run the filter and when.

Updated last run id and timestamp in MS SQL server and added the update logic in Java on load methods. On load methods were triggered when the filter was saved, edited and tested.

2) Added last run id and timestamp to jQuery Ajax requests. Displayed them to the UI using Angular JavaScript. New features were used by hundreds of districts.

1) Implemented deserialization in parameter converter.

2) Wrote unit tests in Spock framework for REST API endpoints.