Marketing, Data Science & Analytics

I help businesses make smarter decisions by blending
marketing insights, data science, and analytics.

Who Am I?

I’m a data enthusiast, passionate about blending marketing, data science, and analytics. Over the past 4+ years I’ve built dashboards that monitor business performances, automated marketing performance tracking across multiple channels, and developed ML-based tools to flag potential risks.

My experience spans agri-tech, e-commerce, and finance-related environments, giving me a unique perspective on how to turn complex data into actionable strategies that improve both customer experience and business performance.

Skills/Tools & Expertise

Technical Skills/Tools

Python 4+ years

SQL & Databases 4+ years

Data Visualization 4+ years

Predictive Modeling 3 years

Appsflyer, Amplitude, Clevertap 3 years

Google Ads, Meta Ads 2+ years

Soft Skills

🧠

Critical Thinking

💬

Stakeholder Management

🎯

Problem Solving

👥

Team Work

🛜

Remote-Ready

🔄

Adaptability

Languages

English Fluent

Mandarin Fluent

German Intermediate

Case Studies

ROAS Attribution Pipeline & Dashboard

Automated revenue-to-campaign attribution across online & offline channels, enabling reliable ROAS calculations for GTM teams.

SQL API Integration Metabase

Value Chain Price Dynamics

Leveraging statistical modeling and machine learning to analyze the complex interplay between pricing, market competition, and demand forecasting across the value chain

Python Scikit-learn XGBoost

Cohort Retention & Growth Accounting

Data-Driven Customer Retention & Growth Optimization

Python SQL Excel

Personal Projects

🏋️

Gym Capacity Tracker

Real-time Web Application

Built a real-time gym occupancy monitoring system with Flask backend, React frontend, and Supabase database. Features QR code check-in/out, live capacity tracking, and analytics dashboard.

Python Flask React Supabase Chart.js

🔗 GitHub 🌐 Live Demo

📊

Frequent Pattern Mining

Statistical Modelling

Applied frequent pattern mining techniques to optimize marketing strategies, enhance product bundling, and improve the overall shopping experience.

Python Jupyter Mlxtend

🔗 GitHub

🤖

Expense Tracking Automation

saving me >90% of time spent logging expenses!

Automated the process of data entry from credit card statements and data labeling of spending categories using GPT models.

n8n Mistral OCR GPT models

🔗 GitHub

Work Experience

Data Analyst

Yara International

May'21 - Jul'25

Drove marketing and product analytics initiatives to empower over 4 Million users and 15K businesses in Africa/Asia's agriculture sector, reducing manual reporting by 50%, lowering online acquisition costs by 40-60%.

Tax Analytics Intern

Inland Revenue Authority of Singapore (IRAS)

Jan'21 - May'21

Predictive modelling to flag out GST Refund Claims and translation of data preparation script from SAS to Python to use on Databricks

Pricing Actuarial Intern

SCOR Reinsurance

Sep'20 - Dec'20

Conducted statistical analysis on Long Term Disability/Critical Illness portfolios based on clients' historical accounts and developed new pricing tools for policy riders (Total Permanent Disability and Backdating)

Education

B.S. Statistics

National University of Singapore

2017 - 2021

Specialized in statistics and data science. Thesis on deep learning for time series forecasting.

Certifications

Machine Learning Specialization

Stanford University & DeepLearning.AI

Comprehensive course covering supervised learning, unsupervised learning, and deep learning fundamentals.

Developing AI Applications with Python & Flask

IBM

Hands-on experience building AI applications using Python and Flask. Project on emotion detection using Flask and Watson NLP library.

Cohort Retention & Growth Accounting

Data-Driven Customer Retention & Growth Optimization

Project Overview

Challenge: Understand user retention patterns and project future growth for planning marketing and product strategies

Approach: Analyzed retention and growth patterns to forecast active users.

Outcome: Delivered a growth accounting model to forecast monthly active users and overall platform growth for business planning.

Actions

Data Cleaning and Visualisation

Conducted cohort analysis from event data to track retention by user acquisition month and country

Retention Analysis

Identified trends in user stickiness and drop-off points across cohorts.

Active User Forecasting

Integrated retention insights into a growth accounting model to forecast monthly active users and overall platform growth.

Key Outcomes

Projected expected user growth over the next 6 months

Scenarios Analysis: Optimistic / Baseline / Pessimistic

Enabled marketing teams to allocate resources more effectively, prioritising high-retention channels.

Visualized in cohort heat maps and retention curves

Tools & Technologies

SQL Python Pandas Matplotlib Seaborn Excel

ROAS Attribution Pipeline & Dashboard

Automated revenue-to-campaign attribution across online and offline channels, enabling reliable ROAS calculations for GTM teams.

Project Overview

Challenge: ROAS reporting was manual, slow, and error-prone due to limited extraction capabilities from Appsflyer.

Approach: Led a cross-functional project to design attribution logic, document data quality, and build an automated dashboard.

Outcome: Delivered a standardised, self-service tool that cut reporting time and improved ROAS data accuracy across regions.

Countries

15+

Active Campaigns

Hours Saved per User per Month

Actions

Project Leadership

Led project timeline and cross-team coordination with engineering to ensure accurate attribution data via API endpoints.

Data Documentation

Created a data dictionary and validated data quality (distribution checks, attribute definitions).

Attribution Logic

Designed multi-touchpoint attribution logic in SQL, capturing the complexity of user journeys across campaigns.

Dashboard Development

Built a Metabase dashboard that links Google Ads, Facebook Ads, Offline Campaigns to revenue

Key Findings

Regionally standardised and automated reporting of campaign performance.

Clear, reusable data documentation adopted by the broader data team.

Replaced restricted 90d Appsflyer exports and manual joins with a single dashboard view, reducing report prep time by ~80% and enabling full-period (multi-month) ROAS analysis.

Empowered GTM teams to self-serve KPI reporting and craft consistent performance narratives.

Tools & Technologies

SQL API Integration Redshift Metabase Mobile Measurement Partner

Value Chain Price Dynamics

Leveraging statistical modeling and machine learning to analyze the complex interplay between pricing, market competition, and demand forecasting across the value chain

Project Overview

Challenge:Multi-tiered supply chain with fragmented data sources provided no visibility of downstream price flow and insights into factors influencing sales price and order volumes across the value chain.

Approach: Analyzed multi-level transactions to uncover actionable insights price dynamics, elasticity, and competitive behavior across distributors, retailers, and farmers.

Outcome:Delivered a predictive model to forecast demand and a statistical framework to determine optimal pricing, leading to more informed business decisions and an expected increase in profitability and sales.

Streamlined
Sales Outreach

40%

Uplift in Repurchase Window Prediction Accuracy

70%

Baseline Model Accuracy

Actions

Data Integration and Preparation

Combined disparate data from sources like SUN ERP system and Redshift data warehouse, setting the blueprint for future data lake integration.

Lag Analysis - Downstream Price Lag & Competitor Lead-Lag

• Calculated downstream price lag analysis to quantify how distributor/retailer price changes propagate to end users.
• Performed competitor lead–lag analysis and causality tests to identify price leadership.

Elasticity Modeling

Conducted price elasticity modeling to understand the elasticities of 4 key product lines

Sales Window Analysis

Simple predictive model on signaling when a distributor will repurchase

Demand Forecasting

Tested XGBoost, Multilinear Regression models to forecast demand volume

Key Findings

Retailers respond to price changes from distributors almost immediately. Developing targeted support strategies to mitigate margin squeeze helps to maintain strong partnerships

Elasticity modeling quantified elastic and inelastic demand for different product lines, providing business confidence in pricing decisions to increase revenue and to grow in market share.

Repurchase signals can be integrated into sales team dashboard to prioritize sales outreach

Segmented distributors based on consistency and volume of purchasing behavior for targeted marketing strategies and data quality monitoring

Established baseline predictive accuracy of demand forecasting model at 70%

Tools & Technologies

SUN ERP Redshift Python SQL Scikit-learn XGBoost Time-Lag Analysis