When a Simple Algorithm will Do

Sometimes when building apps, we come across a simple problem and try to attack it with a very complex solution. We may, for example, envision training a complex neural network to track a customer’s behavior, when all we need is to collect some data and feed it a simple, good old-fashioned algorithm.

In designing custom notifications last summer for a new app, I ran into this exact situation. While the temptation was to run to CoreML, to catch up on the latest and greatest trendy APIs, my ultimate solution, while not anything particularly fancy, ended up being remarkably effective at solving a common problem many apps face: How do you create useful—not annoying—local reminders?

In this talk, we’ll walk through how my app analyzes your logging behavior and custom tailors local notifications to remind you to keep logging. I’ll discuss the design of the app’s notification generator and show how it works in code.

With some thoughtfulness in the design process, our apps can be more intelligent without having to become PhDs in Machine Learning. If you spend more time on algorithm design, rather than flashy new frameworks, you often can build a more pleasant and delightful experience.

Location: Aspen Ballroom Date: August 26, 2019 Time: 2:45 pm - 3:30 pm Joe Cieplinski Joe Cieplinski