active project · 2026–present
Trading Lab Platform
A self-hosted crypto-market research and execution-workflow platform built with explicit boundaries between exploration, paper trading, and live operations.
The idea
Market research can begin with a small script and quickly become a system: historical data, charts, scanners, hypotheses, backtests, portfolios, operational state, and eventually carefully controlled execution.
Trading Lab Platform is the application and runtime side of that system. It is designed to make research reproducible while keeping exploratory work, simulated trading, account inspection, and live actions visibly separate.
What I built
The platform grew from a utility for checking and loading Binance candle data into a FastAPI backend and React workspace. It now covers historical-data maintenance, interactive charts, symbol and pair scanning, category research, exploratory backtests, controlled validation runs, paper portfolios, systematic workflows, and operational reconciliation.
Historical market data is stored in DuckDB. Separate application state records users, workspaces, research lineage, portfolio state, plans, orders, fills, approvals, and operational evidence. Background workers maintain research jobs, history, valuation, and systematic workflows.
The frontend uses typed contracts and includes charting, research, portfolio, and execution-oriented views. Docker packaging and supervised development profiles make the different services and workers reproducible across local and deployed environments.
Safety as architecture
Paper and live environments are separate, and ordinary local development is read-only by default. Live operations require several independent conditions: owner authorization, an explicitly enabled runtime mode, an allowed workspace and credential profile, an active account, policy and risk approval, and a separate execution approval.
Commands and outcomes are persisted so that incomplete or uncertain operations can be reconciled instead of guessed. Secrets are fetched at runtime and kept outside the repository.
These controls are an engineering boundary, not a statement about trading performance. The software is for research and education and does not constitute financial advice.
Current state
Trading Lab Platform is an active private repository. Its source code, deployment, account details, strategy results, and operational interfaces are intentionally not linked here.
The project demonstrates work across data engineering, financial-domain modeling, asynchronous processing, API contracts, interactive visualization, defensive execution design, and operational recovery.
Still to add
I will later add the personal research questions that motivated the platform, privacy-safe screenshots, a narrower walkthrough of one representative workflow, and lessons from evolving a small data utility into a long-running system.