Being the solo engineer for Aeroh has its challenges. I find it tough to keep the full context in mind while I code on several tech stacks. It’s similar to running hundreds of microservices on a development laptop. You will run out of RAM, and if you swap to disk / SSD, it will affect application performance. In my case, my productivity.
But with ChatGPT, things are so much better. ChatGPT with GPT-4 is becoming such an integral part of my development workflow that coding will never be so much fun again without ChatGPT. It has made context-switching so much easier, and my productivity has increased by at least 10x!
For instance, when I switch stacks, let’s say from Android App Development to ESP32 Firmware Development, to implement something, I provide the problem statement to ChatGPT and ask it to come up with an implementation. Usually, ChatGPT does come up with a very close implementation. This acts as a quick refresher and a good starting point. From here, I either use the generated code snippets or ask ChatGPT for specific modifications, and it obliges.
Today, I had three stories at hand. One on ESP32 Firmware, and two on Android App. I broke those stories further down into around seven tasks. This would have taken me around two work days to implement, and I would have probably skipped 2 of the 7 tasks as low priority. But with ChatGPT, it took me close to 4 hours to finish everything, and it was a delightful experience.
I made 16 commits in a 3-4 hour timespan (I usually commit a lot more than an average engineer), but these commits were across three different repositories — Java (Android) , C (ESP32), and Ruby (Rails).
ESP32 (C)
Android (Java)
Ruby on Rails (Ruby)
Before ChatGPT, it would have been multiple searches on Google, with separate queries for API Docs, and example implementations on Stack Overflow, and GitHub with a less than ~1/4 probability of finding relevant answers.