Quantitative Research · Stochastic Modelling & Optimisation · Infrastructure Valuation under Uncertainty
Quantitative researcher. UCL PhD in Systems Modelling and Optimisation, awarded July 2026, specialising in stochastic modelling, numerical optimisation and statistical model validation.
Five years of computational modelling applied to infrastructure valuation — pricing the physical layer that compute and automation run on. Work spans the full model lifecycle: formulation, calibration against observed data, implementation, out-of-sample testing, and interpretation into a decision.
Five peer-reviewed publications and two manuscripts under review; first and corresponding author on five. Journal peer reviewer for Elsevier and IEEE titles.
Currently — Quantitative Researcher at CATL, London: markets, valuation and technology strategy.
| Repository | What it does | Stack |
|---|---|---|
| Uncertainty-Valuation | Two-stage stochastic programme with recourse: non-anticipative first-stage decisions, scenario-dependent second stage, coupled uncertainty sources carried through to a risk–reward distribution rather than a point estimate. | Python · MILP |
| DCResilience | Resilience-oriented sizing for mission-critical load: operating simulation under outage and curtailment conditions, unmet-load and availability accounting, trade-off surfaces between capacity, power and cost. | Python |
| LifecycleValuation | Lifecycle-cost and asset-valuation model — uncertain inputs propagated through state evolution to levelised cost, with parameter checks, bounded error reporting and a reproducible run path. | Python |
| PriceModel | End-to-end price-forecasting pipeline: data ingestion, feature and sequence construction, RNN training and evaluation, split into modular components for reuse and testing. | Python · TensorFlow |
| ML-STOCK-PRICE | Machine-learning study on equity price series: scripted experiments, input data, diagnostic outputs, and a written report documenting method and limitations. | Python · Jupyter |
| Interactive Simulations | Notebooks for scenario analysis, sensitivity testing and communication of dynamic system behaviour to non-specialist reviewers. | Jupyter |
Published as H. Ma and T. Ma; † denotes first and corresponding author.
Peer-reviewed
- T. Ma†, G. Qiao, X. Zhang, D. Hou, C. Spataru, G. Nikiforidis, S. Du. "Degradation-Aware Assessment of Dominant Factors in Performance, Durability, and Cost of Proton Exchange Membrane Water Electrolysers." International Journal of Hydrogen Energy, 264 (2026) 156855.
doi - H. Ma†, G. Nikiforidis, C. Spataru. "System Modelling and Sizing Optimisation of PEM-Integrated Hybrid Energy Storage for Data-Centre Resilience." IET Conference Proceedings, 2025(44), 149–154.
doi - H. Ma†, G. Nikiforidis, S. Du. "Multiscale Modelling and Electrochemical Validation of PEM Electrolyser-Coupled Hybrid Energy Storage Systems." IEEE SPIES, pp. 1–6, 2025.
doi - S. Ishaq, H. Ma, Y. Li, G. Nikiforidis. "Design and Optimisation of Binder-Free rGO/AlO(OH)/Al₂O₃ Aerogels for Energy Storage." Materials Today Sustainability, 31 (2025) 101217.
doi - Y. Li, J. Ren, H. Ma, A. N. Campbell. "Technical and Economic Performance Assessment of Blue Hydrogen Production Using a New Configuration Through Modelling and Simulation." International Journal of Greenhouse Gas Control, 134 (2024) 104112.
doi
Under review
- H. Ma†, C. Spataru, W. Yang, X. Lv, G. Nikiforidis, P. Carvalho, S. Du. "Uncertainty-Aware Optimisation of PEM-Integrated Hybrid Energy Storage for Data Centres: Cost, Carbon, and Resilience Trade-offs." Applied Energy, 2026.
- H. Ma†, G. Nikiforidis, C. Spataru. "Operational Modelling and Resilience-Oriented Sizing of Hybrid Battery–Hydrogen Storage Systems for Mission Critical Data Centres." IET Smart Grid, 2026.
Stochastic simulation at scale — a 20-year Monte Carlo engine at 15-minute resolution, roughly 0.7 million time steps per path, solved repeatedly across scenario sets rather than at a single base case. Solve path restructured to take a multi-day computation down to an error-checked overnight run.
Surrogates that hold up — reduced-order models cross-validated against the full engine to roughly 500× faster revaluation, approximation error quantified and bounded before any result is released. Lifetime operating cost calibrated to within ±3% of observed data.
Optimisation under uncertainty — two-stage stochastic programmes with recourse over 100 coupled scenarios, tail events deliberately retained rather than averaged away. Required model fidelity shown to depend on the target metric: one error measure fell from roughly 8% to 2% under refinement while another stayed near 1%.
Model challenge — out-of-sample robustness and optimality metrics, hypothesis testing with multiplicity control, reconciliation of reduced models against physical benchmarks, to establish whether a conclusion is driven by a real effect, a binding constraint, or a modelling artefact.
| Languages | Python, C++, MATLAB, SQL |
| Python | pandas, NumPy, SciPy, scikit-learn, Matplotlib — modular, object-oriented libraries with unit tests, validation harnesses and version control |
| C++ | CMake builds, Python bindings for compute-bound solver routines |
| Stochastic | Ornstein–Uhlenbeck and mean-reverting calibration, Monte Carlo, scenario construction and k-means reduction, uncertainty propagation |
| Statistics | Hypothesis testing with multiplicity control, out-of-sample validation, goodness-of-fit, VaR / CVaR / expected shortfall, P10/P50/P90 |
| Optimisation | Two-stage stochastic programming with recourse, MILP via PuLP/CBC, dual-based interpretation of binding constraints |
| ML | Surrogate and reduced-order modelling, cross-validation with error bounds, recurrent networks for sequence data, SHAP, TensorFlow/Keras |
Experience
| CATL | Quantitative Researcher — Markets, Valuation & Technology Strategy | London · Nov 2025 – present |
| Huawei European Research Institute | Quantitative Researcher — Simulation & Asset Valuation | Munich · May 2023 – May 2025 |
| Global Energy Interconnection Research Institute (GEIRI) | Quantitative Analyst — Commodity Price Risk & Asset Valuation | Berlin & Birmingham · Oct 2021 – Apr 2023 |
| University College London | Doctoral Researcher — Stochastic Optimisation & Model Validation | London · Nov 2024 – Jul 2026 |
| XPeng Europe | Business Analyst | Amsterdam |
| WMG, University of Warwick | Research Assistant | Coventry |
Education
| PhD, Systems Modelling and Optimisation | University College London | 2024 – 2026 |
| MSc by Research, Systems & Control — Distinction | University of Warwick | 2022 – 2024 |
| Doctoral research, Chemical Engineering & Computer Science — registered; degree not awarded | University of Birmingham | 2021 – 2022 |
| MSc, Sustainable Energy Engineering — Distinction | University of Nottingham | 2019 – 2021 |
| BEng, Energy and Power Engineering — Outstanding Graduate, top 5% | Southeast University | 2015 – 2019 |
PhD thesis: Multiscale Modelling and Optimisation of the PEM Electrolyser–Battery Hybrid Storage System for Data Centre Reliability. Supervisors: Prof. Catalina Spataru, Dr Georgios Nikiforidis.
Founder and convener of Sustainable Horizon Lab, an invitation-only international research network of roughly 90 doctoral-level and above researchers in energy and sustainability, across institutions including the Max Planck Society, Stanford, Chicago, Tsinghua, NTU, TU Munich, Manchester, HKUST, Tokyo and Washington University. Founding member of OYSS.