Dushan Wijesinghe

I lead analysis across customer support operations, from data extraction and wrangling to advanced analytics, providing actionable insights to CX leadership. My work includes applying causal inference techniques (e.g., quasi-experiments, propensity score matching) and predictive analytics to identify drivers

of customer satisfaction and operational efficiency.

Enhanced customer satisfaction levels from 77% to 85% through building and monitoring CS KPIs and escalating customer pain points enriched with data.

First to implement AI in the company with the use of LLM (Large Language Models)(DBRX) for an end-to- end project to improve communication between Customer and Product, which was ideated by myself.

Assisted user research team to understand customer pain-points in one of the products using machine learning techniques i.e. Topic Modelling, more specifically LDA (Latent Dirichlet Allocation) model.

Transformed raw customer support data into structured, business-ready datasets through data modeling, cataloguing, and pipeline automation, enabling reliable analysis across qualitative and quantitative dimensions.

Built a forecasting model using exponential smoothing techniques to help product managers anticipate potential shortfalls in our investor product cash reserves.

Performed comprehensive analysis with the help of TF-IDF analysis using CX data to decide whether the company should provide insurance to our customers through a third party or using internal resources.

Designed and implemented a forecasting framework using LSTM neural networks combined with discrete event simulation to anticipate customer support demand and optimize capacity planning, improving resource

allocation and scalability.

Conducted a factor analysis on employee survey data, combined with random forest models, to identify key drivers of employee satisfaction and performance.

I supported several functional teams in the CS department with data and analytics. I worked on a project with the workforce management team to efficiently forecast FTE requirements when there are unexpected currency delays that happen. They would input the information that they already know, and using historical data, my model would predict the number ofcontacts that specific delay may generate along with a decision matrix. I worked on a project with CS operations to build a unified dashboard to measure quality assurance for all customer support team leads, in which I also introduced the concept ofpower analysis. I worked with the contact handling team to establish KPIs for the newly integrated Chat system into our customer support. I mainly worked with the continuous improvement team to provide forecasting data ofcontacts generated and required FTE distribution for the departmental customer support restructuring that the regional leads were doing. For this purpose, I used the Python library FB Prophet and a prewritten Python script to perform the forecasting. Along with the continuous improvement team, I led a project to understand the usage of our Confluence pages by the customer support agents and developed a metric called "speed to competency". This metric attempts to improve how fast new agents learn and how access to knowledge can benefit them.

Supported and trained new agents to achieve targets and deliver stellar performance.

Moved from specialist to Senior within 1 year