Singapore, Singapore
• Automated 1000+ rows of data entry from banks’ webpages by building a Generative AI application with Python and Google’s Gemini Pro API o Minimized manual market research effort by the business development team and delivered results to be used by senior management for data-driven decisions • Wrote data engineering scripts to move Jira and in-house testing data silos to a new data lake, using Google Cloud Storage and BigQuery o Optimised infrastructure by evaluating configurations and storage and compute trade-offs • Designed and developed a dashboard tool for the finance team to generate invoices to customers and perform analytics using Looker and blockchain transaction data on BigQuery o Engaged stakeholders for 3 rounds of requirements, delivered features planned for next year, closed the project, and handed it to production
• Develop a statistical testing pipeline to measure effectiveness of e-commerce, using Python in Linux environment, following best practices for testing, version control, and CI/CD • Combine data sources: Amazon API, web scraping, and PostgreSQL database for monitoring experiment results, anomaly detection, and hypothesis testing • Evaluate if machine learning models for cost prediction and budget optimization correctly translate business requirements and make improvements in Python on internal AI stack
- Deliver 2 projects in 3 months as the team’s technical lead - Gain practitioner skills in 3 new platforms - Bridge sales operations team with IT experts to design, implement, and review A) Automated Invoice Claims Processing with OCR and NLP • Reduce estimated manual effort by 50% by designing and implementing end-to-end AI software on Dataiku MLOps platform, on top of Hadoop cluster • Utilise Python scripting for unstructured dataset preparation on the Windows command line • Research and evaluate several NLP libraries and PyTorch-based deep-learning OCR model • Deliver a compelling proof of concept: demonstrate an 82% extraction rate of key information and a 95% accuracy in flagging test samples B) Performance Scorecard Dashboard with SAP and Tableau • Automate bi-monthly manual process to daily refresh: Develop workaround process in SAP Business Intelligence reports, eliminating manual editing and Excel queries • Engage non-technical internal customers and incorporate feedback iteratively, ensuring a customisable dashboard which aligns with their needs • Implement vital features, including advanced filters, seamless data connection, and in-built KPI indicators to meet all requirements for a user-friendly dashboard in 6 weeks