5 Tips from an OpenAI Machine Learning Engineer

6 minutes

Everyone wants to talk about AI, more specifically, OpenAI’s GPT, the most advanced large multi-modal model out there. Last week, we had the opportunity to chat with Ted Sanders, a Machine Learning Engineer at OpenAI, whose job it is to explore what GPT is bad at and try to improve it. 

Naturally, we have lots of burning questions about AI. We talked all about how to start a career in AI, why developers still need to learn to code, and the potential of AGI. If you missed the fireside chat, you can watch the whole thing on YouTube below. (We’d love to see you at one of the upcoming community livestreams, so be sure to RSVP.)

Ted is a self-taught developer who got his start coding basic HTML/CSS in Neopets (fun fact: he also used Codecademy). He’s had an impressive career working as a Data Engineer at Netflix and a Physicist at Stanford, but he knows firsthand how daunting breaking into tech can feel. Read on to hear his top tips from the chat for folks who want to launch a career in AI, learn the most practical and applicable AI tools, and get started working with OpenAI’s API today. 

Coding and software development is even more important

A big question on a lot of learners’ minds is: If ChatGPT can essentially code for you, what are the skills worth learning now? The short answer is a little bit of everything. “Coding is only going to become more valuable,” Ted says.

Generative AI is just the latest advancement that enables us to programming computers more efficiently and at higher levels of abstraction. As language models get better, programmers might be able to program well writing fewer lines of code — but that doesn’t mean programming is dead. “In the short-to-medium, maybe even long-term, software development will just become more important and more powerful,” Ted says.

Learn something new for free

As for the specific skills, languages, and frameworks you should learn now? It depends on what you’d like to do. If you want to make apps and websites, learn HTML, CSS, and JavaScript. If you’re interested in machine learning and training models, consider Python. “In general, if you’re trying to become a well-rounded programmer, it’s good to learn a few different languages,” Ted says. “You don’t want to be monolingual, because there are a lot of concepts repeated from one to the next.”

You can add AI to your area of expertise

As more organizations prioritize AI, there are more exciting job opportunities popping up. But the AI sector can feel intimidating, especially for folks who are self-taught developers, career switchers, or people who don’t have advanced degrees. So how can you stand out?

The good news is you don’t need to become the leading expert in AI to distinguish yourself. One way to make yourself valuable in your career is by focusing on getting really good at a few skills and carving out your own niche. “There’s actually a lot less competition in the intersection [of skills],” Ted says. “So if you’re good at sales and AI, now you might be the best AI salesperson in the world. And you’re not really having to compete with the best AI people or the best salespeople, you’re only competing with people at that intersection.”

Curious how you can use AI to distinguish yourself? Read this article to learn how to round out your skill set with AI in a variety of tech careers, and be sure to explore our catalog of AI courses on ChatGPT and generative AI.

Avoid AI tunnel vision

AI is the hottest topic in tech right now, so you should absolutely be well-versed in emerging AI technologies and take courses on AI techniques and topics (BTW, we have a ton of new free AI courses). That said, nobody can really predict the full extent of AI’s impact on developers’ careers. For developers who want to have lasting careers in AI, it’s important to cast a wide net and continue to diversify your skill set.

In many cases, what matters is actually skills complementary to AI, in addition to the skill of using the AI.

Ted Sanders
Machine Learning Engineer at OpenAI

It can be a mistake to throw all your energy into just learning AI, Ted says. Building something cool and useful requires a blend of AI expertise and the ability to create applications and solutions that leverage AI effectively. “In many cases, what matters is actually skills complementary to AI, in addition to the skill of using the AI,” he says.

ChatGPT is a good tool for beginner devs

ChatGPT can boost your productivity as a developer, whether you’re using the tool to debug code or quickly look up a function. And it’s not just the seasoned devs who should take advantage of AI tools; using ChatGPT is clutch for less experienced developers who don’t have “a vast repository of knowledge,” Ted says. “It’s super helpful for just learning the basics, getting up to speed, and asking questions that you might feel embarrassed asking in public or to more experienced coworkers,” he says.

Obviously, there are limits that newbies and professional developers have to keep in mind. For example, ChatGPT isn’t trained on any data past 2021, and the AI chatbot will make stuff up or lie through its (proverbial) teeth. AI can supplement learning to code, but it can’t be a stand-in for hands-on coding.

“I think the best way to teach is with interactivity,” Ted says. “It’s very expensive to have a teacher do it. But if you can have things like Codecademy, things like ChatGPT that can sort of take you along step by step and give you customized feedback, I think it’s super, super useful.”

Get started with the OpenAI API

Part of OpenAI’s mission is to foster open access to its technology. “We want people to build with our technology, we don’t want to keep it to ourselves,” Ted says. With an OpenAI account, you can easily get a developer key to access OpenAI’s API, and “actually build your own applications using ChatGPT’s intelligence to do all sort of different things,” he says. (In fact, our engineers at Codecademy just created a ChatGPT plugin, which you can read about here.)

Before you jump in, Ted recommends reading the documentation on the OpenAI website. Also check out the OpenAI Cookbook repository on GitHub, which he used to run. There, you can find lots of code examples and snippets, plus solves for common pain points like counting tokens, and handling rate limits. The Cookbook also has some inspiring examples of how you can use the API for things like vector databases and information lookup. We also have a free course Intro to OpenAI GPT API, which will help you examine the API and create more reliable outputs from large language models.

Feeling empowered to start learning about ChatGPT, AI, and programming? Explore our catalog of AI courses and case studies, brush up on coding basics, and discover new ways of thinking.

Related courses

6 courses

Related articles

7 articles