Experience

• Principal Research Fellow, School of Engineering, RMIT University - 02/2025-Now

  • Responsibilities:
    • Lead AI research across multiple industrial sectors (renewable energy, manufacturing and banking, etc.)
    • Lead research in Large Language Models, Knowledge Representation Learning, Cognitive Agents, Embodied AI, along with their applications

• Director/Principal Research Scientist, IT Innovation and Research Center, Technical Director of AI-TMG, Huawei Technologies - 04/2023-01/2025

  • Responsibilities:
    • Lead 30+ scientists and engineers in advanced IT R&D in Perception, Cognition, Decision Labs
    • Lead innovation initiatives and research collaboration in Huawei IT to support workforce, customer service, IT operations leveraging large language models
    • Lead Huawei IT AI-TMG (AI Technology Management Group) developing AI strategy and technical standard
    • Lead research in Generative AI, developing data processing and enhancement pipeline for AIGC
    • Lead technology strategic/business planning (TSP/TBP) and technology road-maps designing
  • Achievements:
    • Establishment of Huawei-Beijing University Innovation Lab in advanced speech and language semantics (华为-北大语音语义创新实验室)
    • 2024 Technology roadmaps for Huawei IT System Platform (HIS)
    • Research POCs (LLM Knowledge Dialogue HR System, BabelTar Summarization, Speaker Extraction, Fault Root-cause Analysis) commercialized in Huawei Information System (HIS) Products: HR IT helpdesk, Welink, WeOps, etc.
    • Text summarization wins No. 1 position at Allen Institute for AI GENIE Summarization XSUM leaderboard
    • A strong track record of research outcomes published in top AI conferences and journals (11 in NLP, 4 in Multi-modal, 3 in CV, 1 in Cybersecurity)

• Director/Principal AI Scientist, AI Application Research Center, Product Manager of Artificial General Intelligence, Huawei Technologies — 01/2021–04/2023

  • Responsibilities:
    • Lead 50+ scientists and engineers in research and development of AI products/algorithms in NLP/CV/Data Science/Multidisciplinary AI
    • Lead innovation initiatives and research collaboration in Huawei IT
    • Lead Huawei IT AI-TMG (AI Technology Management Group) in developing AI strategy and technical standard
    • Develop research culture and in charge of talent hire
  • Achievements:
    • Strong research partnerships with top-tier institutions (Tsinghua, Beijing University, HIT etc.) with a track record of research publications in top conferences and journals (23 publications with 16 in NLP, 3 in CV, 1 in AI in medicine, 3 patents)
    • Generative AI SDKs with billions calls yearly covering Manufacturing (OCR/Face Recognition/Defect Detection), Customer Service AI (Dialog/KGQA), Conference AI (Speech Machine Translation/Speaker Diarization), etc.
    • Lead team to achieve rate of research transfer 50% (research papers and patents)
    • Develop Huawei AI application blue book, AI standard and strategy plan
    • Prestigious young talents hire
    • Win WMT 2021 biomedical translation tasks (No. 1 place for EN→FR, ZH→EN, EN↔IT)

• Principal NLP Scientist, AI Application Research Center, Huawei Technologies — 01/2019–12/2020

  • Responsibilities:
    • Lead developing neural machine translation systems (NMT) for low resource language
    • Lead domain adaptation research for NMT
    • Lead data augmentation (sentence alignment, dictionary-enhanced) research for NMT
    • Lead developing multi-turn, multi-domain dialog system
    • Lead research in dialogue management methods (DST, Dialog Policy)
  • Achievements:
    • NMT models deployed to Huawei HMS and HIS with BLEU scores for 20 language pairs outperforming Bing Translator
    • BLEU scores of domain-specific NMT outperforming Google Translator
    • Win WMT 2020 biomedical translation tasks (No. 1 place for EN↔FR, EN→DE, EN→IT; 2nd place for EN→RU, DE→EN, RU→EN, IT→EN)
    • Win WMT 2020 parallel corpus filtering tasks (No. 1 place for Pashto, No. 2 for Khmer)
    • Win WMT 2019 biomedical translation tasks (No. 1 place for EN↔FR, EN↔DE; 2nd place for EN↔ZH)
    • A track record of research results published in top conferences and journals (8 in NLP, 5 in AI in medicine, 2 patents)

• Senior Lecturer in Business Analytics, Research Centre for Data Analytics and Cognition, La Trobe University — 08/2016–12/2018

  • Responsibilities:
    • Perform research in cognitive deep learning, apply data science, artificial intelligence to health care (collaborated with Prof. Mingguang He from University of Melbourne), social media analysis in service (with Prof. John Gero from UNC), smart grids (with Prof. Xinghuo Yu from RMIT University);
    • Develop domestic and international collaborative research group focusing on deep learning;
    • Lead research project to develop constructive dynamic memory networks in natural Language understanding (with La Trobe University Research Start-up Grant);
    • Lead international collaborative research to develop AI-based recommendation algorithms for Chinese Medicine (with Tianjin Chase Sun Pharmaceutical);
    • Lead industry mobile computing project aiming to enhance effectiveness, efficiency of warehouse picking (with Shiny Pty Ltd);
    • Lead advanced analytics project aiming to enhance customer experience (with Shiny);
    • Teach and coordinate subjects of Principle of Business Analytics, Customer and Social Media Analytics and Analytics Industrial Internship Capstone projects;
    • Co-supervise PhD candidates in applied artificial intelligence and data analytics
  • Achievements:
    • High quality publications: 1 edited book, 4 journal papers, 2 conference papers
    • Program Co-chair for the 30th Australasian Joint Conference on Artificial Intelligence
    • Student feedback 4.4 out of 5 (with mode 5) for Master of Business Analytics Subjects

• Senior Data Scientist at Big Data Analytics, CTO & Innovation, Telstra — 04/2015–08/2016

  • Responsibilities:
    • Develop customer experience score board and predictive model for fixed broadband customers (Hive, Qlik sense, Random Forest in h2o)
    • Develop video systems for face and object recognition (OpenCV, Scikit-learn SVM)
    • Lead developing text sentiment engine using machine learning approaches for NPS text Verbatim (NB Multinomial, SVM from R)
    • Develop customer cross sell predictive models based on live chat text data, click streams and demographic data (algorithms NB Multinomial, GBM from h2o)
    • Responsible for co-developing complaints prediction models with imbalanced data (GBM, Random Forest, C4.5 from h2o, RWeka from R)
    • Prepare Kaggle data mining competition project proposal relating post-paid mobile customer insights/experience
    • Sourcing big data for SAS Telstra Hackathon (customer journey data from big data lake using Hive, SAS VA)
    • Supervise junior data scientists (PhD data science interns)
    • Contribute to maintain data analytic server and tools (on R multicore servers) for data science group
  • Achievements:
    • Text sentiment engine prototype with accuracy above 92% in real time testing
    • Cross-sell predictive model on live chat text with AUC 0.8
    • Enhanced complaint prediction model with +20% recall rate over benchmark model
    • Enhanced cross-sell prediction with lift of 6.6 for top 1% population over benchmark
    • Fixed broadband customer interaction prediction model with 0.80 AUC

• Technical Lead Data Scientist at Global Analytics Hub, Lenovo — 01/2014–04/2015

  • Responsibilities:
    • Responsible for setting-up advanced big data analytic platform, managing junior data scientist/engineer, presenting/demo to global leadership/senior executives
    • In charge of developing advanced data analytic tools for social marketing purpose including customer segmentation/loyalty program from social media (Twitter, 微博) using machine learning and NLP algorithms (Latent Dirichlet Allocation)
    • Lead developing context-based sentiment analysis and components categorisation to discover customer and product insights from unstructured data
    • Lead developing topic trends detection to guide contents generation and maximise social media customer engagements
    • Lead developing ontology, dictionary on interesting topics and applied them to theme detection for social media (twitter) data
    • Support customer-centric product design (Laptop, Smartphone features) using advanced statistics – conjoint analysis to produce optimal product features
    • Support brand channels insights from marketing research data using statistic hypothesis testing
    • Support social media listening for customer insights deep-dive using text mining (Topic Modelling)
    • Support global corporate strategy via providing advanced data-driven solution
  • Achievements:
    • Social media customer segmentation/classification engine that is used to automatically identify Lenovo target customer segments from social media based on text contents (in R tm, topicmodels, openNLP, shiny, RWeka, etc.) – demo available
    • Social emerging trend engine based on twitter, Google hot trends (in R shiny, twitteR, RCurl) – demo available
    • Customer insights to support Lenovo internal stakeholders

• Senior Data Mining Analyst/Acting Manager at Australian Transaction Reports and Analysis Centre, AUSTRAC — 09/2010-12/2013

  • Responsibilities:
    • Leading data science projects providing insights/metrics/monitoring for financial transactional data
    • Chief Investigator for near-real-time funds flow prediction via developing time series forecast models to identify abnormal financial transactions
    • Chief Investigator for entity behaviour pattern recognition via developing text mining models
    • Chief Investigator for funds flow analysis using multivariate statistics and social network analysis to identify significant entities and relationships
    • Chief Investigator for developing metrics for gauging effects of cross agency projects through behaviour analysis
    • Hand-on data extraction/processing/mining on heterogeneous structured and unstructured data source
    • Data analytic report writing and outcome presentations
  • Achievements:
    • Data mining/machine learning and analytic systems for financial data
    • Research reports provide solutions to impact on national strategic and operational decision-making

• Research Fellow at School of Electrical and Computer Engineering, RMIT University — 07/2008–09/2010

  • Responsibilities:
    • Lead data analytic research project in developing smart energy application in short-term load/peak demand prediction for electricity using statistic algorithm (ARIMA, ETS, ANN)
    • Apply artificial intelligent (multi-agents) methods to simulate end-use scenarios of household electricity loads accommodating historical weather data and appliance models (NetLogo)
    • Develop middle-term, long-term load forecasting methods using multivariate statistics with end-use scenarios and historical weather data
    • Lead research project in developing artificial intelligence (Abductive Reasoning, OWL Ontology) models for geometric reasoning in engineering design
    • Coordinate local group seminar in information system engineering lab and provide data mining/analysis training to the team of researchers
    • Liaise with industrial, academic collaborators to initiate projects in energy, water resource and health informatics areas
  • Achievements:
    • A Multiple-run Interactive Certainty Network in explanation generation (implemented in JAVA)
    • ARC-linkage grant proposals in smart adverse drug interaction detection system and SNOMED concept learning for drug-disease precaution early detection and refinement
    • High quality publications: 1 book, 1 journal paper, 1 book chapter, 5 conference papers

• Research Scientist at Commonwealth Scientific and Industrial Research Organisation (CSIRO), Tasmanian ICT Centre — 02/2007–07/2008

  • Responsibilities:
    • Develop artificial intelligence stochastic model simulating/analysing household demand data for Energy Transformed Flagship (Smart Grid)
    • Develop battery coordination algorithm (by introducing new battery coordination algorithms and testing new capability) to optimise distributed generation
    • Design and implement advanced data management/AI platform for hydrology simulation and intelligent irrigation control
    • Design and implement machine learning algorithms for soil profile classification
    • Co-author journal, conference papers and project reports
  • Achievements:
    • New features for CSIRO patented “StigSpace” Simulator (in Matlab)
    • Stigspace Simulator Software Benchmark report, Stigspace Simulator Software Description report
    • Journal publication in IEEE intelligent systems

• Part-time Research Associate at Key Centre of Design Computing and Cognition, University of Sydney — 03/2003–12/2006

  • Responsibilities:
    • Situated Agents, Artificial Intelligence projects with CRC-CI
    • Develop UI, rule schema for DesignCheck™ using Java, Express-X
    • System integration with CAD model, EDM database and report system
    • Develop data mining algorithms for smart building asset management in AIMM™ (Agents in Maintenance Management)
    • Co-author academic papers and projects reports
  • Achievements:
    • A commercialized software system DesignCheck™ (in Java) which is a knowledge based system automatically check CAD design compliance to Australian building code
    • A prototype system package AIMM™ (Agents in Maintenance Management)