As a ML engineer with over four years of experience in the AI field, I’ve witnessed the remarkable evolution of this space. Among the many innovations that have emerged in recent years, ChatGPT stands out as one of the best. It’s like having a super human assistant right at your fingertips! People have been using ChatGPT for various tasks, such as managing their diets or planning travel itineraries. In this blog post, I’ll share my personal experience of using ChatGPT for the last six months and and how this is helping me in my job.
Outlines
- The denial mode
- Experiment Documentation
- In AI Research and ML Project
- Picking up new skill
- DevOps
- Writing — Blog, Email
- Conclusion
The denial mode
When ChatGPT was first released, I thought it might be similar to other text generation models like Bert or GPT and that it was all just hype. But after spending some time understanding it and interacting with this, I was amazed by its abilities! In just a few days, ChatGPT became a tool I use every day. It’s been incredibly helpful and impressive.
Experiment Documentation
As a machine learning engineer, there are times when you need to conduct many experiments in just a few weeks. It can be tiring to record all these experiments result and observations. So what I do is basically put in all the details using simple bullet points. Then, I provide this information to ChatGPT with prompt on how I want it organized. ChatGPT does that job easily. And now you have saved easy 10 minutes!
In AI Research and ML Project
I use AI to solve other AI issues, and it’s amazing how advanced our world has become. Seriously, it’s true!
Right now, I’m working on my own project called NoiseBlock. It helps analyze audio from online meetings and gets rid of the annoying background noise as we speak. When I began this project, I knew nothing about audio & noise cancelation or where to begin. ChatGPT really helped me speed up my progress.
At another instance, I had to create a custom model for an image classification problem. It involved using a CNN-based architecture, and the feature maps were passed through FPN (Feature Pyramid Network) before the classification step. If I had to manually write the whole architecture, it would have taken me a lot of time, maybe even an entire day! With the help of ChatGPT, I managed to finish it within just an hour.
So yeah long live ChatGPT ;)
Picking up new skill
As an ML Engineer, you’ll always need to learn new skills, whether it’s for your side projects or your job. ChatGPT makes it incredibly convenient to acquire new skills quickly.s. Let me share two instances where ChatGPT came to my rescue, saving me a lot of time.
- Working with Go Language for the first time — I encountered a key service codebase written in Go Lang and needed to make some minor fixes for a specific edge case. Being a noob at Go language, it would have taken me quite a while to figure it out on my own. However, with the assistance of ChatGPT, I was able to grasp the basics, understand the codebase, and start development within just two days.
- Learning SwiftUI for macOS app development — As mentioned earlier, I’m currently working on NoiseBlock, which will be a macOS app. Coming from a pure ML background with little knowledge of building direct consumer-facing UIs, ChatGPT proved immensely helpful in teaching me Swift, SwiftUI, and how to create apps for a broader audience.
DevOps or Cloud
Many ML professionals often lack cloud skills due to their primary focus on ML tasks. However, it’s important to understand cloud concepts, especially when working on a product that handles a substantial number of requests. Personally, I’ve turned to ChatGPT several times to help me understand various cloud-related topics.
Once, I had a situation where I needed to move from CPU-based deployment to GPU to improve response times. However, I faced a problem with the pod scale-up time. While CPU pods took around 1 minute to scale up, GPU pods were taking much longer, around 3–5 minutes. Thankfully, ChatGPT helped once again. I simply asked how to reduce GPU pod scale-up time, and I got to know about image streaming in Kubernetes, which significantly reduced the pod scale-up time.
Writing — Blog, Email
Content writing appears to be one of the most popular applications of ChatGPT, and I also use it for the same purpose. When it comes to writing blogs or emails, I find ChatGPT helpful in the following ways:
- Ideation and Research: ChatGPT helps me in generating ideas and conducting research for my writing projects.
- Email Formulation: Since I write a lot of emails, I often provide pointers to ChatGPT, and it helps me compose the email with the desired intent, tone, and content.
- Re-writing: When working on an article in a Google document, I go to ChatGPT for help in enhancing vocabulary and overall content improvement, making the article even better. For example I took help from ChatGPT to re-write this pointer as well.
If you enjoyed this article, your applause would be greatly appreciated!
Conclusion
In conclusion, ChatGPT proves to be beneficial to everyone in various ways, becoming an indispensable tool in many people’s lives. As a Machine Learning Engineer, I find ChatGPT extremely valuable in my work.
