Uses
My day-to-day toolkit for building production AI systems: React + Ant Design on the front, Python microservices for the brain, and an observability stack that keeps me honest.
Design
- Figma for quick flows, wireframes, and component specs.
- Excalidraw for sketching data flows and pipelines.
- Lucidchart for higher-fidelity diagrams and process maps.
Development
- React + Ant Design for fast, accessible UI with theming.
- styled-components for design-token-driven styling.
- Vercel for deploys, previews, and edge/CDN.
- Cloudflare Tunnel to safely expose dev services without public IPs.
- Puppeteer (Node/TypeScript) for web automation when APIs don’t exist.
- Node.js for small API glue, workers, and build tooling.
Data & ML (Python)
- Poetry for packaging and reproducible envs; CLIs with Typer.
- pandas + SciPy for analysis; Numba for accelerating hot paths.
- Property-based tests with Hypothesis.
- Connectors/backtests against Binance Spot API (with strict controls).
Infra & Observability
- InfluxDB OSS for time-series storage.
- Metrics with Prometheus, dashboards in Grafana, logs via Loki.
- OpenTelemetry (Python) + structlog for structured logs/traces.
- Docker Compose for local orchestration.
System
| Desktop | Windows 10 workstation (primary dev box) |
|---|---|
| Operating system | Windows 10 |
| Browser | Google Chrome (stable) |
| IDE | PyCharm for Python; editor varies for JS/TS. |
| Laptop | Desktop-first workflow |
| Audio | Over-ear + in-ear (varies by session) |