+ Forecasting information cascades in social networks: Designed novel approaches to summarize and estimate the statistics and evolution of information spread in social networks based on partial observations in an accurate and scalable manner.+ Mining discriminative subgraphs from global-state networks: Proposed a fast and accurate algorithm to extract discriminative subgraphs that best predict theglobal state of networks given local states on nodes.+ Answering top-k representative queries on graph databases: Proposed novel index structure to efficiently compute top-k answer set that is both relevant and representative in graph databases with speedup of up to three orders of magnitude over baselines.

Project: Spatio-temporal model for predicting urban traffic flows from taxi GPS data.+ Devised a scalable spatio-temporal algorithm to predict the flows of taxis and bikes for different functional regions of Beijing and New York with over 40% improvement over baselines.+ Collaborated with 2 team members on collecting and cleaning business data.

Project: Expertise-based clustering of artists on Behance.net online social network+ Designed an algorithm to detect communities of creative users in a large heterogeneous network of 7 million users, utilizing both their profiles and their activities to capture users' expertise.+ Worked biweekly with newly acquired Adobe company to ensure smooth transfer of data and relevant business-related project goals.+ Presented research findings and reported experimental results biweekly on a new community detection algorithm to other team members and the lead researcher of the acquired company.

Project: Distributed system for analyzing call data records+ Designed and implemented a Hadoop system to detect anomalies in phone call networks.+ Proposed a new method for detecting anomalies in evolving networks.+ Discussed weekly with 2 IBM researchers on anomaly detection algorithms and system design.

Project: Structural properties and evolution of collaboration networks: A simplicial perspective+ Studied group dynamics in the formation and evolution of collaboration networks.+ Discussed weekly with a BBN researcher on evaluation metrics for group dynamics, and presentedfindings to researchers at the US Army Research lab.

Projects: Friend suggestion and feed ranking systems+ Designed and implemented Friend Suggestion System for Zing Me, the largest Vietnamese social network with over 10 million users, and increased friendship suggestion accuracy from 53% to 86%.+ Designed and implemented Feed Ranking System using Redis for storage and Thrift for interfacewith web-based applications.+ Communicated with various business departments (e.g., front-end and back-end teams) to implementefficient system design and smooth product launches.

Project: Biometric-based network access control and information security system+ Researched multimodal biometric fusion at matching score level, leading to an improvement in error rate from 2.57% to 0.17% in comparison with single-fingerprint system.+ ImplementedandintegratedmultimodalbiometricauthenticationsolutionintohybridBiometric/PKI-based access control system.+ Collected fingerprint data from 50 students at HUST to test the new authentication solution in a realistic environment.

+ Researched security vulnerabilities (SQL-injection, heap-based and stack-based buffer overflow). • Analyzed malicious codes and software vulnerabilities; issued patches and fixes.+ Implemented security software, including local network and software vulnerability scanners.+ Trained new interns on popular tools and techniques related to security vulnerabilities.