Atomic Cache
Years of market data on your own disks, with a dashboard that shows what is cached, what is running and what needs a look.
About
Overview
Atomic Cache keeps a trading team's market data on its own storage: minute bars from 2016, end-of-day option chains and intraday option quotes for every symbol it tracks. Research and trading tools read from the cache instead of the data vendor, so a backtest that needs years of history gets it from disk, with no rate limit in the way.
The dashboard shows how much of the tracked universe each store covers, how fresh it is, and which backfills are running or queued. Adding a symbol takes one form. The request is queued, a background worker pulls the history, builds the derived timeframes and reports progress step by step.
A nightly check looks for gaps, short histories and unexplained price jumps. Routine fixes run on their own and are logged; anything that needs judgement waits on the dashboard for a decision. Scheduled jobs, service health and per-minute API traffic sit on one page, so a missed run is visible the next morning.
Built with Python, FastAPI and PostgreSQL, with the data stored as Parquet files and a Next.js dashboard. Other services read the cache through HTTP APIs, and AI agents get a generated guide to every store and how to request more.
Media
Screens & stills
Highlights
What makes it work
Years of history on local disk
Minute bars from 2016, end-of-day option chains and intraday option quotes, stored as Parquet and read in place by research and trading tools.
Add a symbol, watch it fill
One form queues the backfill. A worker pulls the history, builds every timeframe and shows progress for each step, with an estimate for the queue behind it.
Data quality checked nightly
Gaps, short histories and unexplained price jumps are found each night. Routine fixes run on their own; anything that needs judgement waits on the dashboard.
Every dependency on one page
Service health, scheduled jobs and per-minute API traffic, so a missed run or a slow endpoint shows up before anyone relies on the data.
Have a project in mind?
Tell us what you're planning: a website, a game or a video. We'll reply with how we'd approach it, a rough timeline and a cost estimate.