• Productionised and maintained end-to-end location-based Division Autocomplete API Service utilising ElasticSearch and Flask (QPS: 20, P95 Latency: 15ms), expanding service into 2 countries• Developed highlight feature for Division Autocomplete using ElasticSearch Fast Vector Highlighter, improving P95 latency from 100ms to 15ms

Part of the (MAP) POI Search Team dealing with NLP analysis of Shopee user address over multiple countries - Indonesia, Vietnam, Malaysia, • Engineered a Big Data ETL Pyspark pipeline conducting feature engineering for over 600m rows of messy data into a centralized database used in Shopee’s core services in Vietnam• Designed an airflow process to automate manual reruns of ETL pipeline, saving 15 hours of work monthly• Constructed complex ElasticSearch matching queries improving Geocoding accuracy from 91% to 96%• Drove business decisions by conducting geospatial analysis using QGIS and KeplerGL

• Performed data collection and feature engineering to craft datasets for graph network visualisation & analysis with Gephi, Python & R

• Coached weekly sessions to clarify students’ doubts on fundamentals of Business Analytics• Performed grading of student assignments

• Built a Natural Language Processing (NLP) Web application scraping and visualising sentiment scores of news data from top 30 Nasdaq companies - communicating analytical insights to various audiences• Employed React & Django with pre-trained Google Finbert, a pre-trained model designed to analyse sentiment of financial text

• Developed an automated Java & Python pipeline for large graph visualisation using Gephi Java Toolkit and DataShader to monitor and measure trends across the Ethereum network• Authored 2 analytical articles for marketing outreach on cryptocurrency hacks (Upbit Hack) and on growth of Stable Coins requiring business thinking, analytical insights and customer centric visualisations• Pitched and garnered keen interest from 2 new clients globally (Monetary Authority of Singapore & Nexo)

• Joint Authorship for Manuscript accepted into PLOS Computational Biology• Conducted data cleaning, visualisation and analysis on multi-dimensional (American Gut Microbiome) abundance data with over 730 prediction variables to clean and process data• Engineered an Expectation-Maximisation algorithm package to iteratively infer microbiome interactions from microbiome abundance and external perturbation data• Designed and developed a simulation pipeline for model testing, attained an accuracy of up to 97%