- Scope, analyze and create data contract to extract granular information for segmented capacity planning requirements for over 500 data centers

- Develop and deploy a strategic capacity plan tool for global data center servers (pods), space and power planning for over 15 data centers, with an annual opex lease spend of over $500 million.- Automate manual capacity planning tool that resides in spreadsheet, replacing it with a system developed using relational database as backend data storage and Python as planning engine. The tool enables strategic what-if analysis of servers (pods) space/power requirement and financial impact due to different input changes (such as customer data storage savings, server hardware density improvement, server age life span). - Create and maintain Tableau dashboards to communicate 5-year planning plan of record to downstream space/power procurement teams and financial partners.- Create long range forecasts using Facebook’s Prophet model for regional servers(pods) compute and storage usage over a 5-year horizon and track the forecast performance via tableau visualization.- Analyze and quantify different hardware generation’s compute performances using quantile regression to create a statistical confidence range on the point estimate forecast.

- Develop and deploy a strategic quantitative analysis tool for global AWS data center space and power planning for over 25 clusters and 75 availability zone (AZ), impacting the global data center annual capex spend of over $2 billion. The tool enable regional planner to perform various scenario analysis to access the space/power requirement and financial impact of new region launch, existing region long lead time assets (e.g. land, shells, substations) acquisition and new AZ launch cost benefit analysis. - Create and maintain Redshift data tables and views related to regional long range capacity planning for downstream business partners (finance/ real estates). The views are then produced into various data visualization metrics for state of the region space and power analysis.- Served as a SME in working together with SDE team to bring the quantitative model from Excel VBA platform into a full scale production model. - Develop and deploy a driver-based planning model to transform customer usage demands into server racks power requirement for long range planning. The model enable a plan-to-plan variance analysis that enable bridging of key output (such as power requirement) to various contributing components of the inputs. - Create and maintain data pipelines to programmatically extract data from disparate data sources into S3 buckets, transform and load the data into Redshift clusters for downstream regional planning

Design and develop world wide strategic customer service agent resource planning tool, that enable the process of operational and financial planning, what-if scenario analysis, and plan-to-actual/plan bridging for the global capacity planning team.- Developed an algorithm in PL/SQL in order to implement a strategy that enable filling up low cost internal customer service sites, while keeping outsourcing rate constant and each internal site to be at certain fill level, in order to capitalize the lower cost structure within the internal sites. The strategy resulted in more than $7 million annualized savings.- Developed a 4 year outsourcing strategic analysis that optimizes the variable cost per contact. The analysis utilized a linear program to optimize the cost while keeping certain constraints in place, such as site capacity fill rate, cultural affinity to the customer service agent to the customer, and quality score. The estimate annual savings ranges between $15 to $20 million under different assumptions.