Quantitative Researcher & Engineer · End-to-End Quantitative Systems
I research, build, operate, monitor, and evaluate quantitative systems end to end. My work sits at the intersection of markets, statistics, engineering, software, and applied AI.
Founder of Qinvia · Dubai, UAE
LinkedIn · X · Medium · Qinvia Research
- Quantitative research: systematic strategies, performance measurement, falsification studies, and portfolio/risk analysis.
- Research infrastructure: market-data pipelines, audit tools, APIs, reproducible notebooks, and production-oriented Python systems.
- Applied AI: research workflows and human-first software where AI improves the process without obscuring evidence or judgment.
Research question → data contract → implementation → validation → deployment → monitoring and review
I work across the full cycle. Research is only useful when the assumptions are traceable, the software is reliable, the result survives costs, and the live system can be monitored and challenged after deployment.
| Project | Evidence and purpose |
|---|---|
| ETF Universe | Bilingual study of ETF exposure, intentional management, survivorship limits, financing assumptions, and performance versus SPY, with reproducible notebooks and frozen evidence. |
| Pairs Trading: TRUMP & MELANIA | Visual, cost-aware introduction to OLS, Kalman hedge ratios, spread signals, and conflicting cointegration diagnostics using real Binance perpetual-futures data. |
| Factor-Residual Reversion | Cost-aware falsification study across 50 liquid OKX perpetual swaps. The pre-cost effect is measurable; the locked cost test is negative, and the full result is published. |
| Relative-Wealth Martin | Cash-relative path-efficiency metric with a Python implementation, bilingual notebooks, 1,967-ETF study, and transparent out-of-sample validation. |
| Direction–Breadth Factor | Experimental signed direction and effective-magnitude-breadth descriptor, with algebraic tests, bilingual research notebooks, and a DOI-backed release. |
| Beyond Sharpe: TRACE & Coherence | Auditable implementations of experimental path, drawdown, concentration, and direction–breadth diagnostics, with their limitations stated explicitly. |
| Project | What it provides |
|---|---|
| OKX Quant Toolkit | Read-only collection of public OKX candles, trades, funding, books, open interest, derivatives data, and bounded WebSocket captures. |
| Binance Quant Toolkit | Research-first access to public Binance market data, archive planning, provenance, order-book reconstruction, and data-quality checks. |
| Qinvia Return Series Auditor | Dependency-free CLI that exposes data-quality defects and statistical fragility in dated returns and equity curves. |
Evidence before narrative. Costs before claims. Limitations beside results.
I prefer bounded questions, explicit data contracts, reproducible code, and results that remain useful when the original hypothesis fails. Negative results are publishable results when the method and stopping decision are clear.
The public repositories hold code, notebooks, tests, data contracts, and derived evidence. Long-form editorial editions live on Qinvia Research so GitHub remains a technical, inspectable record rather than a second website.
My background combines engineering, quantitative finance, and software development. I have presented quantitative work publicly and built systems across research and production contexts.
Python · C++ · Rust · SQL · pandas · PyTorch · TensorFlow · AWS
The tool is secondary to the research contract: make the question clear, the inputs traceable, the implementation testable, and the conclusion proportionate to the evidence.
Qinvia — quantitative research and systems development.
qinvia.com