Curriculum vitae
As a data scientist, I build machine learning solutions using PyTorch / Keras / scikit-learn, and I also develop optimization models with Gurobi / PuLP / CPLEX. My past work spans HVAC energy optimization, video analytics, text-to-SQL generation, predictive modeling, anomaly detection, physics-guided machine learning, survival analysis, and schedule optimization in logistic industry.
In parallel, I contribute to the LIGO Scientific Collaboration, applying Bayesian inference to black-hole astrophysics. My research focuses on parameter estimation of black-hole progenitors, particularly within the pair-instability mass gap. Recently I had the opportunity to speak at PyCon Hong Kong’s 10th anniversary.
Resume
Downloadable PDF (Industry): PDF
Downloadable PDF (Academia): PDF
Education
Research experience
- Aug 2021 - Present: Research Assistant (Part-time), Universidad de Santiago de Compostela
- Supervisor: Professor Juan Calderón Bustillo
- Research Collaboration: LIGO Scientific Collaboration
- Researched on parental paramter estimates of black holes (BHs) in GW190521:
- Developed a Bayesian framework to compute posterior probability distributions for the ancestral properties of the component BHs in GW190521. The framework is applicable to any hierarchical formation scenario.
- Investigated the hierarchical formation of BHs within the pair-instability supernova mass gap, analyzing their masses, spins, and birth recoils.
- Evaluated scenarios in which the merger components were retained in their environments, finding that if GW190521 was quasi-circular, a nuclear star cluster origin is plausible for \(p_{2g} \in (∼0.4, ∼0.8)\).
- Optimized and revamped the research group’s numerical simulation codebase, achieving up to 6× runtime performance.
- Released the designed framework as an open-source package archeo (10.5281/zenodo.14569049) and deployed a frontend for preview at Streamlit Community Cloud.
- Co-authored “Kicking Time Back in Black Hole Mergers: Ancestral Masses, Spins, Birth Recoils, and Hierarchical-formation Viability of GW190521” (Carlos Araújo-Álvarez et al 2024 ApJ 977 220).
- Contributed as an analysis team member and co-authored “GW231123: a Binary Black Hole Merger with Total Mass 190-265 \(M_{\odot}\)” (A. G. Abac et al 2025 ApJL 993 L25).
- Contributed as an analysis team member and co-authored “GW241011 and GW241110: Exploring Binary Formation and Fundamental Physics with Asymmetric, High-spin Black Hole Coalescences” (A. G. Abac et al 2025 ApJL 993 L21).
- Researched on Hubble Constant measurement with intermediate-mass BH mergers:
- Explored the feasibility of measuring the Hubble Constant using gravitational-wave (GW) signals detected by LISA.
- Simulated GW waveforms of intermediate-mass BH mergers and analyzed conditions to break mass-redshift degeneracy.
- Constructed redshifted waveform pairs and computed signal-to-noise ratios to identify redshift drift in year-long signals.
- Feb 2025 - Aug 2025: Research Assistant (Part-time), Johns Hopkins University
- Supervisor: Professor Emanuele Berti
- Research on deep-learning-based parameter estimation of post-merger GW signals.
- Sep 2020 - Dec 2021: Student Researcher (Part-time), The Chinese University of Hong Kong
- Supervisor: Professor Tjonnie G. F. Li
- Research Collaboration: LIGO Scientific Collaboration
- Researched on searches for strongly lensed GW images:
- Conducted GW data analysis and parameter estimation using bilby and PyCBC.
- Simulated 200 lensed signal pairs under different signal-to-noise ratio settings, and generated sky localization probability maps (skymaps).
- Developed and evaluated overlap statistics for filtering non-lensed pairs, achieving >99% filtering efficiency at a false positive rate of \(10^{-2}\).
- Co-authored “Using overlap of sky localization probability maps for filtering potentially lensed pairs of gravitational-wave signals” (2112.05932).
- Jun 2021 - Aug 2021: Research Intern, The Chinese University of Hong Kong
- Supervisor: Professor Kenny C. Y. Ng
- Researched on semi-analytical model of solar atmospheric gamma rays with Potential-Field Source-Surface (PFSS) Model
- Developed semi-analytical methods and numerical simulations to reproduce solar atmospheric gamma-ray flux.
- Utilized HMI synoptic maps to compute the global coronal magnetic field and simulate the trajectories of relativistic cosmic-ray protons.
- Shortened simulation time from weeks to hours, enabling efficient calculation of emission probabilities and total gamma-ray flux.
- May 2020 - Aug 2020: Research Intern, The Chinese University of Hong Kong
- Supervisor: Professor Tjonnie G. F. Li
- Researched on searches for strongly lensed GW images (see above).
Work experience
- Sep 2025 - Present: Data Scientist, IBM Hong Kong
- Build Monte Carlo simulations to model disruptions and identify potential breakpoints in aircraft schedules.
- Develop stochastic optimization models delivering up to 50% delay reduction in schedule across 100+ aircrafts.
- Coordinate with cross-functional teams to gather requirements, design solutions and data pipelines, and deploy models into production environments.
- Jun 2024 - Sep 2025: Associate Data Scientist (non-trainee), OOCL
- Researched on program-aided reasoning for advanced table understanding with LLMs.
- Initialized and developed common library for optimization to:
- allow quick switch between multiple LP frameworks including CPlex, PuLP, pyomo, and Gurobi,
- standardize the development of LP models across different projects,
- support sensitivity analysis and functionality like parallel solve of optimization problems.
- Temporarily led a data science project for 3 months:
- planned tasks for 4 data science team members and supervised a data scientist trainee,
- handled production issues and troubleshooting,
- communicated with developer teams for data pipeline development and managed deployment schedule.
- Applied clean architecture to revamp an optimization engine for bunker procurement planning, achieving up to 2× runtime performance improvement.
- Designed bunker procurement optimization algorithms, successfully replacing the production model and delivering up to $5 million in annual savings (in pilot scope).
- Conducted sensitivity analysis and backtesting to evaluate the optimization engine’s performance.
- Developed data patching algorithms for missing data imputation and validation, significantly enhanced data integrity from below 50% to 97%.
- Developed Spark workflows for dataset preparation and scheduled optimization.
- Developed Jenkins deployment pipelines for RESTful API services and a web-based chatbot application.
- Designed API and data schema for bunker procurement optimization engine.
- Standardized the team’s development workflow by unifying linting and coding standards, introducing pre-commit hooks, and migrating dependency management to Poetry/uv.
- Set up GitLab CI for automated testing and Docker image building, structured package release processes.
- Researched on bunker price forecasting using machine learning (ML) models and time series analysis:
- Conducted a literature review on bunker procurement strategies and state-of-the-art ML models for handling prediction delays.
- Trained and fine-tuned ML models for bunker price prediction, achieving 2% mean absolute percentage error (MAPE) for short-term forecasts and 6% MAPE for long-term forecasts.
- Experimented with feature engineering and model selection to improve generalization across different bunker types and regions.
- Aug 2023 - Jun 2024: Assistant R&D Engineer, ATAL Engineering Group
- Reference Letter: PDF
- Applied survival analysis techniques and designed a physics-guided model to predict equipment failure and enhance system reliability.
- Revamped HVAC energy optimization system with event-driven structure and reorganized optimization strategies to reduce electricity consumption by 20%.
- Developed a web-based interface for model fitting and visualization of cooling system performance.
- Implemented a RESTful API for facility management and equipment monitoring, integrating it with AWS cloud services and a data lakehouse.
- Designed database schemas and contributed to the MLOps pipeline, streamlining dataset preparation, model training, evaluation, and versioning.
- Conducted a proof of concept to demonstrate automatic SQL generation with large language model (LLM) agent:
- Built a knowledge base for SQL generation, incorporating data schemas, query templates, and SQL examples.
- Developed a self-correction loop to improve SQL query precision, leveraging error feedback from database engines such as Amazon Athena.
- Explored existing LLM models for SQL generation and prompt engineering to enhance SQL generation accuracy.
- Developed a web-based interface for SQL generation and chatbot communication with the LLM agent.
- Jun 2022 - Aug 2023: Software Engineer, Sebit Company Limited
- Reference Letter: PDF
- Researched on sensor data analysis for elevator abnormalities:
- Developed fault detection algorithms using frequency domain and spectral analysis, combined with statistical techniques, to identify elevator abnormalities.
- Implemented a data pipeline for sensor data preprocessing and analysis.
- Researched the use of generative adversarial networks (GANs) to simulate detected abnormalities for training data augmentation.
- Key findings were presented at the 14th Symposium on Lift & Escalator Technologies in the United Kingdom.
- Co-authored “Condition-based and Predictive Maintenance Strategy for Lift Installations using Big Data Analytics” (Proceedings of the 14th Symposium on Lift & Escalator Technologies).
- Developed deep learning and computer vision applications:
- Developed deep learning models for real-time object detection, classification, and Cantonese voice recognition.
- Finetuned state-of-the-art computer vision neural networks for object classification and enhanced the accuracy by 5%, achieving 97% accuracy.
- Implemented tracking algorithms for video analytics and optimized system performance for real-time processing.
- Designed critical detection mechanisms for a lift door monitoring system, which is presented at the 48th International Exhibition of Inventions in Geneva and won a sliver medal.
- Developed model training and deployment workflows using Ansible and Jenkins.
- Dec 2021 - Jan 2022: AI Developer Intern, Flying Milk Tea Limited
- Developed programs for data scraping, data annotation and preprocessing with Blender.
Skills
- Major Tech Stack
- IDE: Visual Studio Code
- Language: Python / SQL
- Tools: Docker / poetry / git / pylint / mypy / black / isort / pytest / pre-commit / PySpark
- Deep Learning: PyTorch Lightning / PyTorch / Keras
- Other Technical Skills
- Language: C++ / C / Go / Bash / R / JavaScript
- DevOps: Podman / ansible / Jenkins / Helm / GitHub Actions
- AWS: Athena / Redshift / SageMaker / VPC / EMR / Lightsail / Bedrock
Publications
LIGO Scientific Collaboration, Virgo Collaboration, & KAGRA Collaboration. (2025). GW241011 and GW241110: Exploring Binary Formation and Fundamental Physics with Asymmetric, High-spin Black Hole Coalescences. Astrophysical Journal Letters, 993(1), L21.
Al-Shammari, S. et al. (2025). GW231123: A binary black hole merger with total mass 190–265 M⊙. The Astrophysical Journal Letters, 993(1).
Araújo-Álvarez, C., Wong, H. W., Liu, A., & Bustillo, J. C. (2024). Kicking Time Back in Black Hole Mergers: Ancestral Masses, Spins, Birth Recoils, and Hierarchical-formation Viability of GW190521. The Astrophysical Journal, 977(2), 220.
Chan, J. K., Leung, C. K., Wong, W. T., Kwok, S. C., & Wong, H. W. (2023, September). Condition-based and Predictive Maintenance Strategy for Lift Installations using Big Data Analytics. In 14th Symposium on Lift & Escalator Technologies (Vol. 14, No. 1, pp. 33-45).
Wong, H. W., Chan, L. W., Wong, I. C., Lo, R. K., & Li, T. G. (2021). Using overlap of sky localization probability maps for filtering potentially lensed pairs of gravitational-wave signals. arXiv preprint arXiv:2112.05932.
Talks
June 28, 2025
Hong Kong Python User Group Workshop at Auki Labs, Hong Kong, Hong Kong
November 16, 2024
PyCon Hong Kong 2024 at City University of Hong Kong, Hong Kong
October 22, 2024
Hong Kong Python User Group Meetup at Oursky Limited, Hong Kong
April 04, 2024
2024 American Physical Society April Meeting at SAFE Credit Union Convention Center, United States
April 26, 2023
48th International Exhibition of Inventions Geneva at Palexpo, Switzerland
September 26, 2020
Annual Physics Student Conference 2020 at The Chinese University of Hong Kong, Hong Kong
Certifications
- Google Project Management Professional Certificate, Google Career Certificates, Jan 2025
- Generative AI for Data Scientists Specialization, IBM, Jun 2024
- Generative AI for Software Developers Specialization, IBM, Jun 2024
- Reinforcement Learning Specialization, University of Alberta & Alberta Machine Intelligence Institute, Mar 2024
- Data Science for Investment Professionals Specialization, CFA Institute, Sep 2023
- Accelerated Computer Science Fundamentals, University of Illinois at Urbana-Champaign, Dec 2022
- TensorFlow Developer Professional Certificate, DeepLearning.AI, Nov 2022
- Google Data Analytics Professional Certificate, Google Career Certificates, Nov 2022
- Japanese-Language Proficiency Test (JLPT N1), The Japan Foundation, Jul 2022
Service
- Jan 2021 - Present: Undergraduate Member, LIGO Scientific Collaboration
- Aug 2023 - Present: Associate Financial Technologist, Institute of Financial Technologists of Asia
- Dec 2024 - Jan 2025: Meetup Host, Open Source Hong Kong and Hong Kong Python User Group
- Jun 2024 - Nov 2024: Proposal Reviewer and Sprint Technical Support, PyCon Hong Kong
