Writing · LinkedIn · Resume · Email
I'm Sachin, based in Singapore. I build production Rust and Python systems at Tamira Technologies and research systematic trading signals. My goal is to bring that combination of software engineering, statistical reasoning, and experimentation into the quant industry.
Education: Certificate in Quantitative Finance, 84/100; B.Sc. Data Science & AI, NTU, Honours (Merit).
Competition: WorldQuant BRAIN Gold League (Top 50), 2023 NTU x WorldQuant Alphathon.
SYSTEMATIC STRATEGIES · Python / pandas / NumPy / statsmodels
Momentum and reversal research across 25 cryptocurrencies. Tests lagged signals, transaction costs, train/validation splits, and walk-forward portfolio combinations, with explicit discussion of survivorship bias and multiple testing.
Research takeaway: individual in-sample winners often fade in validation; diversified walk-forward combinations are the more defensible finding. This is exploratory research with documented limitations.
DERIVATIVES & RISK · Python / Panel / Bokeh
European option pricing with Black-Scholes, analytic Greeks, and spot/volatility/rate scenarios. Exposes a Python API, CLI, and interactive dashboard, with reference-value tests for prices and sensitivities.
Inspect the pricing model · Read the tests
EVENT STUDIES · Python / Polars
Tools for studying market reactions around CPI, FOMC, and earnings announcements. Compares pre/post returns, volatility changes, drawdowns, and time to peak move across assets, with configurable event windows and CSV exports.
FIRST-AUTHOR PUBLICATION · ICDM-SENTIRE 2025 / IEEE workshop proceedings
A retrieval architecture combining stateful neural memory, vector retrieval, and reasoning compression. Evaluated across three datasets, including scientific-literature question answering.
Read the paper · Research code · Python library
- Rust execution infrastructure: maintain and extend a production execution router at Tamira, with a 3x latency reduction and ownership of configuration and deployment.
- Reproducible evaluation: improved semantic-routing accuracy from 67.5% to 89.3% across 6,102 benchmark cases through retrieval diagnostics, ranking, and contextual disambiguation.
Earlier work spans physics-informed and transformer-based time-series modelling at NTU, multi-agent software systems at MAOTO, 3D perception and multimodal planning at Schaeffler, and production RAG and retrieval systems at Dell.
- parquet-zed - Rust tooling for inspecting, previewing, and editing Parquet data through a CLI and Zed integration.
- xmonad-miral - An experimental Wayland compositor connecting a Haskell window-management model to a C++/MirAL host.
Signal robustness after costs. Portfolio construction. Derivatives and risk. Market microstructure. Research infrastructure that makes experiments easier to reproduce and inspect.
LANGUAGES
SCIENTIFIC COMPUTING
DATA & INFRASTRUCTURE
Away from code: chess, retro operating systems, and breaking.
Public activity. Language shares reflect repository size, not proficiency.
Happy to connect about quant research, research engineering, and open-source tooling. Get in touch.