AI/ML Solution Architect with a proven track record of designing and implementing cutting-edge AI, ML and Generative AI solutions. Proficient in ML Engineering and System Design, with expertise in Machine Learning, AI Risk, Data and Cloud technologies.
Work experience
2021 Present
AI/ML Solution Architect
AIA Singapore
-
Solution Architecture Leadership:
- Orchestrated the design and implementation of sophisticated data and ML solutions tackling challenges such as Real-time Lead Allocation and Recommendation Systems, catering to a daily user base exceeding 1.5 million, driving a significant business impact of S$1M.
-
Cutting-Edge Technology Adoption:
- Led the integration of latest advancements in Data Engineering and ML Engineering domains, leveraging technologies like MLFlow Pipelines, Metaflow, Delta Tables, and Serverless Spark Streaming jobs, ensuring the delivery of highly efficient solutions.
-
Proof of Concept and Innovation:
- Conducted comprehensive proof of concept evaluations for Generative AI applications, utilizing LangChain and OpenAI models for Text Analytics, including tasks such as Text Summarization, Q&A, and Chatbot development.
-
Model Development and Governance:
- Engineered the Distributor Fraud and Conduct model, facilitating proactive monitoring and reporting of potential Financial Mis-selling activities by SG AIA Agents.
- Implemented robust AI Governance strategies, incorporating ResponsibleAI and Explainable AI practices, to ensure compliance with MAS regulations across Banking and Finance projects.
-
Infrastructure Migration and Vendor Management:
- Oversaw the seamless migration from Oracle to Databricks for data analytics and machine learning pipelines, ensuring minimal disruption and maximum efficiency.
- Served as the key liaison for interfacing with Databricks vendors on ML Use Cases, fostering continuous engagement to expedite cloud adoption efforts.
-
Operational Efficiency and Impact Measurement:
- Formulated and executed MLOps and DataOps strategies, driving end-to-end automation for 10 models within the team, optimizing operational efficiency and scalability.
- Collaborated on various ML use cases, including supervised classification for people analytics and unsupervised user segmentation for marketing, meticulously quantifying the impact of each use case.
-
Strategic Analysis and Decision Support:
- Conducted insightful strategic analyses leading to revisions in business strategies and processes, leveraging causal analysis techniques to provide data-driven decision support and enhance organizational effectiveness.
2020 2021
Data & ML - Solutions Architect
Kepler FI
- AlphaScale trading system was designed to ingest trading signals from multiple trading strategies(from over 10+ researchers) and to generate alpha over standard indices like S&P.
- Responsible for solution architecture and design for adoption of trading system
- Built the C4 style architecture for the ML pipeline and defended the architectural decisions to a panel of architects
- Designed and built a centralised low latency caching data structure for high speed access to large datasets. Reduced time to load from a few hours to a few mins.
- Leveraged AWS infrastructure to support Machine Learning Research and introduced software best practices to the workflow
- Partner with production systems team to analyze production incidents, determine corrective action, and find improvement areas to better support operations
- Worked with business process owner to identify improvement opportunities
- Back-testing of the trading system on the portfolios of the parent firm resulted in +ve alpha leading to adoption of AlphaScale as a strategy for the parent trading firm
20192020
Senior Machine Learning Engineer
Grab | Grab Financial Group
- Developed extremely precise gibberish text detection models, with a precision and recall of 96% and 88% improving manual reviews to human in the loop AI system.
- This model was also eventually integrated into GrabDefence, an enterprise fraud protection offering from Grab.
- Developed fraud detection models to reduce chargebacks from customers leading to millions of dollars in savings and improved user experience.
- Worked with other ML teams to develop a strategy for addressing the class imbalance problem utilising data-programming techniques resulting in improved model precision and recall across the different markets(SG, PH& ID)
- Quantified model impacts using counter factual evaluation and pass-through for fraud models.
- Worked along with other team members on developing PoC for automated KYC using image detection and OCR techniques.
20172019
Machine Learning Developer
SAP
- A hybrid learning recommendation system built using deep learning approaches that is now used by over 3000+ enterprises the likes of which include Disney, HSBC etc.
- Implemented model explainability tools that reason about the models prediction results
- Researched and implemented algorithms to solve unsupervised topic matching
- Conducted testing of the model predictions to understand the model performance in different scenarios leading to a better understanding of model limitations. Built metric reports to evaluate the performance of the ML models in production
20152017
Machine Learning Engineer
PropertyGuru
- Developed and deployed the unit price prediction model to estimate the market price of a housing unit. The first such estimate developed by a private firm in Singapore
- Built the entire machine learning pipeline from scratch, from extracting raw data to retraining the model and batch inferencing on a daily basis. Monitor the model performance using Tableau dashboards
- Developed economic indicators such as price trends, listing demand, median price, transaction volume, etc across different geographic boundaries of Singapore
- Developed the entire application from building indicators, to exposing the results through an API response to display the price trends
- Built a state of the art data warehouse solution on GoogleCloudPlatform
- Participated in formulating and designing the data-lake from scratch
- Developed end-to-end data pipeline from data ingestion to building data marts that were consumed by Tableau dashboards
- Developed scheduling systems to monitor job flows
20102013
Programmer Analyst
Cognizant Technology Solutions
- Created and maintained business reports on top of large enterprise datawarehouse and the reports were consumed by the Investment Banking units of JPMC, NY.
- Built interactive scorecards and dashboards using JavaScript and jQuery, providing customisation not readily available through reporting tools
- Was awarded excellence in delivery by Global Head Enterprise Information Management(JPMC)
Education
2014 2015
Master of Technology
National University of Singapore
Knowledge Engineering
2006 2010
Bachelor of Technology
Kumaraguru College of Technology, Anna University
Information Technology