Web application & automation
Resy Reservations
An application that monitored restaurant availability in New York and made reservations matching the client’s preferences.
- Preferences
Restaurant, party size, dates and times.
- Monitoring
Continuous checks of the booking page.
- Availability
A table within the selected windows.
- Reservation
Booking through the signed-in browser.
This page draws on my experience building the project. The original materials are no longer available, and the diagram is illustrative.
Finding a table before the slot disappeared
The client lived in New York and wanted to book tables at restaurants where reservations filled as soon as they became available. Checking the pages manually meant being there at just the right moment.
I built an application where he could choose a restaurant, party size and acceptable date and time windows. The system then monitored availability on Resy and made a reservation when it found a matching option.
From the interface to the server
I developed the React frontend, the Python backend and the browser automation with Selenium. I also deployed the application to a VPS and configured Nginx as its reverse proxy.
Monitoring ran on the server. The interface collected the client’s preferences, while the backend coordinated page checks and the steps needed to complete a reservation.
Keeping the browser running and signed in
Maintaining an authenticated session was one of the harder parts. The automation needed access to the client’s account to complete a booking, and I used session cookies to preserve that access between interactions. Selenium located page elements using XPath selectors.
Another issue appeared on the VPS: in that environment, the site did not load correctly in a headless browser. I configured a virtual display on the server so the browser could run with a graphical interface without a physical monitor.
These adjustments were part of operating the application. Finding a table also required keeping a browser that could navigate the booking flow.
Use and retirement
The client used the bot at least once a week for much of a year.
After Resy changed its interface, the automation needed an update. The client chose to stop using it. Because the integration relied on page elements, changes to the platform required revisiting the selectors and navigation flow.
Technologies
Python · React · Selenium · Nginx · VPS