Now you can Instantly Chat with Parveen!
- 14+ years of experience in analysis, design, development, testing, maintenance and deployment of software systems.
- Experienced in Application development, Web development, OO programming and Client-Server system development. ...
Swift performs deterministic and non-deterministic forecast calculations for pension and retiree welfare plans. Enhanced Swift engine/functionality by implementing multiple tasks i.e. Import customization with rebuild, Swift Compare functionality, channel splits, stochastic splits etc.Show More Show Less
The Scheduler is the master scheduling process in charge of monitoring and orchestrating the flow of scheduled jobs for all of DSS applications. Users can schedule all processes that manipulate or access Data Ware House data from this one point. Scheduler server can execute any number of executable programs. When a process is to be run in a non-interactive mode (batch), it may be submitted to the Scheduler as a job. Jobs are grouped into streams. Such jobs might include data loads and transfers, summarizations and reports, queries, stored procedures, or scripts. Job control information is stored in the Schedule file which needs to be updated with new or updated job information in each release. C# based Scheduler conversion tool performs that task. Enhanced the tool to Replace the old model streams with new models depending on three conditions, one to one, split or merge. Custom jobs in old streams were also moved accordingly.Show More Show Less
Swift performs forecast calculations for pension and retiree welfare plans. It can reflect accounting and contribution policies that correspond to a variety of calculation standards. Swift can be used to perform both deterministic and stochastic forecasts. It is also used as the source of calculations for the Channel, a tool through which we can provide clients the ability to monitor funded status and perform forecasts themselves.
Implementing Stochastic Calculation on Azure HPC Cluster
Swift azure stochastic calculation is a hybrid application. The Azure HPC cluster executes swift Excel workbook computations in parallel with minimum latency,maximum performance and ensure that these are within acceptable bounds for reasonable user experience. The results of individual computations are written to an Azure Sql Database via calls to a Custom Azure WebAPI through workbook custom macro calls. The process also communicates results back to an on premise oracle database via an on premise xml web service.
- Enhanced swift engine to enable stochastic calculation through the azure cluster. Due to the unique nature of Microsoft HPC service for excel and excel macro framework, Implemented C# API to submit Web requests to the stochastic WebAPI through excel HPC macro framework
- Implemented a Stochastic Data Service Web API to interface with SQL Server Database on azure
- Implemented User interfaces for integration with azure
- Data Base Updates including creating/fine tuning Stored Procedures, tables, functions etc.
- Swift HPC cluster property optimization for optimal cluster usages like the amount of max and min resources, auto grow shrink, Spin up required nodes on demand, optimal node size /number, use of broker node etc. for optimizing performance and reduce cost.
- Swift HPC cluster is based on a queued model, Users experience a delay due to queue warm-up or waiting for required resource. Implemented Scheduler rest API utilizing Microsoft HPC Scheduler Rest API interface to provide periodic updates on nodes availability, queue progress etc to maintain a reasonable balance between cost and desire.
- Swift has a separate charting utility that pulls both raw data and post-calculation stats from the database. Implemented an updated charting tool to enable creating Charts using Azure HPC Data.
- Implemented a WCF exposed MSMQ within the current workflow for a reliable, asynchronous communication between swift stochastic HPC client and server to enable multiple user run calculation concurrently in a failsafe manner.
- Analyzed Azure VM performance and storage structure and suggested suitable VM/Storage options to meet the Performance, throughput and scalability requirement of the most demanding stochastic simulations for investment modeling.