I work on temporal representation learning for structured time-series data. My research studies how temporal signals can be represented, decomposed, compared, and analysed to support reliable modelling, scientific interpretation, and reusable research infrastructure.
My broader interests include symbolic regression and temporal structure in large sequential models, world models, embodied AI, and real-world variable-rich systems.
| Project | What it demonstrates | Stack |
|---|---|---|
| EchoWave | Explainable structural similarity for time series and datasets, with plain-English summaries, HTML reports, and compact JSON. | Python, similarity, reports |
| AgentForecast | Multi-backend, agent-friendly forecasting that exports publishable charts, cards, Markdown, and JSON. | Python, forecasting, agents |
| De-Time | Unified time-series decomposition software with one Python and CLI interface for trends, oscillations, residuals, components, metadata, and machine-facing workflows. | Python, decomposition, forecasting |
| TimeSeriesCounterfactuals | Counterfactual workflow tooling for before/after time-series questions, including charts, reports, share packages, and agent-ready handoffs. | Python, causal impact, synthetic control |