Career Profile
Data Scientist, Software Architect, Big Data Engineer in Operational Research, Cloud Solution, Big Data, Optimization, Statistical Inference, Agent-Based Modelling and Simulation, Machine Learning. Backend architect and full stack engineer.
Experiences
Work within the NSI-Science team at the Amazon Selection and Catalog System
Design and implement web and mobile e-commerce solution using Spring Cloud, Magento, react-native, react, angularjs, websocket, elasticsearch
System architect responsible for the design and implementation of CLEF web applications and its distributed computing platform built using Spring Cloud, Spring Boot, Apache Spark + GraphX, Hadoop, Apache Mesos, Mariadb Cluster.
Managing Director Chief technology architect of Meme Analytics Pte Ltd, specializing in data analytics, machine learning, simulation and optimization, web and mobile application and works on intelligent decision support, analytics and optimization system for supporting large scale real world engineering problems in dealing with complex problem dynamics through modeling and simulation, offer consultancy or direct customized system implementation and integration of turnkey solutions to clients.
Data scientist of neuro10, specializing in big-data machine learning for log and timeseries analytics, responsible for designing and implementing big-data machine learning infrastructure as well as algorithms related to anomaly detection, text mining, natural language processing, time series predictive modeling. Actively involved in the development of Java-based software solution using Apache Spark, Mesos, Akka, Cassandra, Kafka, as well as HDFS (Hadoop Distributed File System).
Researcher working on machine learning and agent-based simulations, actively involved in various industrial projects related to Machine Learning, Data Mining, Supply Chain Optimization, Agent-Based Modeling, etc.
Researcher working with NTU, main area of research includes meta-heuristics, continuous and combinatorial optimization, memetic computation, vehicle routing, supply chain risk management. Research work involves genetic programming, multi-objective optimization, robust search, time series forecasting, agent-based modeling, neural networks, gaussian process, PID and MPC controller
Software developer and technical consultant at CodeZone, specializing in delivering custom solutions for scheduling and optimization as well as simulation applications.
Software Developer for Office Tool Automation using VBA. The office tool automation includes code generator that dynamically generate and inject VBA codes into excel based on instructions retrieve from database.
Software Developer for Office Tool Automation using VBA as well as developing Web 2.0 application using ASP.NET. The office tool automation includes automatic processing data from various data sources such as Excel, Access and process results into auto-generated Powerpoint presentation and auto send via Outlook automation. The Customer Satisfaction Dashboard is a Web App I designed and implemented to automatically retrieves and process data on customer satisfaction and send individuallycustomized report to various level of management, based on their roles and tasks involved
Inventions
Journal Publications
Conference Publications
Open Source Contributions
For the video demo of some of the projects i have done in my professional works:
For my personal Open Source contribution, they can be found at the following links:
Projects
Designed and implemented big-data infrastructure for a rule-based data analytics platform with real-time association rule mining for job matching and company ranking
- AI implemented: big data expert system shell based on spark, real-time distributed association rule miner
- Programming Languages: Java, Scala, Javascript
- Technologies used: Azure, Spark, Redis, Hadoop, MariaDB cluster, Nginx, snort, Spring framework, Spring Cloud, Angular, gulp, Mesos, Ansible, Linux shell scripts and Powershell automation scripts.
Designed and implemented real user monitoring system that keeps tracks of user’s real-time interaction with the web application and behavior analytics
- Programming Languages: Java, Javascript
- Technologies used: Graphite, nodejs, Spring Boot, AngularJS, WebSocket
Designed and implemented intrusion detection and prevent system which provides webview, as well as email and chat notification, for both network-based and host-based intrusion detection
- Programming Languages: Java, Javascript
- Technologies used: snort, iptables, nodejs, Ansible.
Implement order preserved encryption system for mariadb to provide both encryption in motion and at rest as well as support for crypto versioning
- Programming Languages: Java
- Technologies used: Spring Data JPA, MariaDB.
Develop a desktop tool that allows user to analyse data from Excel, CSV, or Database using various machine learning algorithms.
- AI implemented: Decision Tree, Genetic Programming, Clustering, SVM, ANN, GLM, MA
- Programming Languages: C#
- Technologies used: Telerik, Winforms, D3
Design and develop large-scale distributed agent-based modelling and simulation platfor for labour market using Akka.NET, provides visualization via 3D simulation and modelling using Unity3d, and desktop + web visualization
- AI implemented: distributed agent-based modelling and simulation based on distributed computational framework
- Programming Languages: C#
- Technologies used: Akka.NET, Consul, MS SQL Compact, Telerik, Winforms, Nancy, ASP.NET, Unity3d
Designed and implemented big-data machine learning solution in the areas of log analytics and time series prediction, the work involves designing and implementing anomaly detection, text mining, natural language processing, time series predictive modeling, graph ming
- AI implemented: topic modelling (online LDA implementation on both Spark and Akka), Word Colocation, NER, Document Categorizer, Word2Vec, Time Series Forecasting, Non-parametric Regression Analysis, Document Clustering, Isolation Forest Anomly Detection, LOF Anomaly Detection, Deep FeedForward NeuralNet
- Programming Languages: Java, Scala, Javascript
- Technologies used: Spark, GraphX, Redis, Zookeeper, Hadoop, Kafka, Cassandra, ElasticSearch, Akka, Spring framework, Angular, nodejs, DeepLearning4j, Mesos, Ansible, Shell scripts
Designed and implemented iOS and Android mobile application using OCR (Optical Character Recognition) and Edge Detection Canny Algorithm for detecting objects in video as well as texts in the video, implemented fuzzy text matching for corrupted context obtained from OCR AI implemented: association rule minining, FPGrowth, sequential association pattern mining
- AI implemented: association rule minining, FPGrowth, sequential association pattern mining
- Objective-C, Java, PHP, Javascript
- Technologies used: iOS, Android, AngularJS, JQuery, XCode, JavaFX, Spring framework (Spring MVC, Spring Data JPA, Spring Security, Jersey), Bootstrap, Jasper Reporting and Dynamic Report, tesseract, OpenCV
Develop the intelligent network management application for network analysis and anomaly detection. Current supported features
- Agorithms implemented: 1. Centrality Measurement, including Eigen Vector Centrality, Closeness, Betweenness, Degree 2. Community Finding: Walktrap algorithm 3. Link Anomaly Detection algorithm 4. Community-based Graph Differential Anomaly Visualization 5. Data Capture and Monitor via Fluke’s Network Time Machine. 6. Graph Models 7. Simulation and Monitoring
- Programming Languages: C#
- Technologies used: Fluke’s Network Time Machine, PostgreSQL, Windows Presentation Framework (WPF), Telerik, MySQL, Windows Communication Framework (WCF)
Develop intelligent discrete event multi-agent simulator for an intelligent multi-agent control for simulating power consumption and occupant comfort in smart building engineering. The system includes space location modelling using surface extraction from planar graph algorithm; discrete event simulation for emulating the IO devices and power units in the building; simulation model for energy consumption modelling; model and measurement for occupant comfort; parameter sweeping features for allow users to simulate various scenarios in multi-dimension feature spaces; GUI user interaction including settings for power utilization and occupant comfort, floor plan design; library/script interface for user interaction and external modules to be built on top of the simulator; graph-based visualization for the state transition in Markov Chain Monte Carlo simulation; unit testing and logging module. The system is also a home automation kit which is interfaced with electrical and other home appliances as well as sensor units for real-time monitoring and control in home automation.
- Algorithm implemented: ARMA, NSGA II, surface extraction from planar graph
- Programming Languages: C#
- Technologies used: Windows Presentation Framework (WPF), Telerik, Windows Communication Framework (WCF), Raspberry PI, emonCMS, X10
Designed and implement a framework and platform to solve various vehicle routing problem (VRP) variants, with 16 solvers including memetic algorithm, genetic algorithm, hyper heuristic, simulated annealing, constructive heuristics, etc, each of which capable of solving one or more (combined hybrid) of the vehicle routing problems such as capacitated VRP, VRP with time windows, VRP with backhaul, VRP with pickup and delivery, multiple depot VRP, VRP with stochastic demands, heterogeneous VRP, as well as dynamic VRP with time windows, with support for routing on map based viewer, parallel processing, simulation, and algorithm performance analytic. The system is designed to be both standalone desktop and enterprise web service, with client in various programming languages (Web Browser, Excel, Java, C#, etc) as the online Vehicle Routing Scheduler Library solvers.
- Algorithms implemented: memeplex-based memetic computing, hyper-heuristics, constructive heuristics, trajectory-based optimization (tabu search and simulated annealling)
- Programming Languages: C#
- Technologies used: Winforms, GMap.NET
Design and develop race tracking and reporting solution from scratch. The software solution allows recording participant race results in real time via connection to IPICO lite and elite readers which forms a WAN with a command center via master-slave network structure. The timing data is processed and streamed to web site in real-time for participants of the race to view.
- Algorithms implemented: Huffman compression, priority queue based on binary heap, left leaning red black binary search tree
- Programming Languages: C#, PHP
- Technologies used: Winforms, GMap.NET
Developed a national strategy analysis and proposition using machine learning approach via Genetic Programming and Evolutionary Algorithms as well as data analytics and prediction such as time series prediction algorithms such as ANN, RBF and ARMA.
- Algorithms implemented: Kmeans clusering, genetic programming, genetic algorithm, ann predictive, ARMA, RBF
- Programming Languages: C#, Javascript
- Technologies used: Winforms, WebKit
Design and implements a polygon mesh processing system that allows users to sculpt polygon mesh in real-time, the system successfully implements the self-adaptive topology for freely sculpting 3D meshes, with several additional improvements, which includes integration with other remeshing and model repairing algorithms, managing the multiple unorganized cross-sections as well as developing offline filter version of the system for its integration with other polygon mesh processing tool.
- Algorithms implemented: Self-adaptive topology algorithm, remeshing algorithm, mesh processing techniques
- Programming Languages: C++
- Technologies used: GLUT, GLEE, OpenGL