Artificial intelligence (also known as AI, but you knew that) is shaping the next phase of technology for everyone. Developers and engineers are already discovering how and where AI fits into their tech stacks. Companies like Apple, Microsoft, and Google are weaving AI models throughout various platforms and services to force users deeper into ecosystems.
For tech professionals, AI can present exciting opportunities. Though technologies like machine learning and natural language processing have been around for quite some time, the new generation of AI tools and services allows them to be used in interesting ways that present new pathways to success and career advancement. Or to put it another way: AI allows tech pros to do their jobs faster and “smarter” than ever before, provided they deploy and utilize this technology correctly.
But those interested in AI might need help figuring out where to start. Keeping up with an AI ecosystem that’s evolving quickly takes a lot of work. Which programming languages are essential? Will a certification in AI help you land a job? Should you contribute to open-source AI repos to stand out?
We spoke to experienced technologists to better understand where to focus your efforts if you want to work with AI in 2025 and beyond.
What should someone interested in becoming an AI developer focus on in school?
As you matriculate, Koushik Sundar, Vice President at Citibank, advises students to focus on the basics: “Building a strong foundation is important, especially in mathematics, probability, and statistics. AI is a broad field that includes deep learning, neural networks, and GenAI. What's evolving now are hybrid solutions that combine multiple tools and algorithms to develop AI agents and tools. To get a holistic view and understand how things work behind the scenes, it's crucial to have a solid foundation.”
Charlie Clark, former senior software engineer at Squarespace and founder of Liinks, agrees.
“I think building a solid trifecta of skills is crucial: math, algorithms, and curiosity for the unknown,” Clark says. “AI, at its core, is about turning data into insights, and that means you need a deep understanding of linear algebra, statistics, and calculus—not to memorize formulas but to know how models think under the hood. Couple this with data structures and algorithms, and you can efficiently build systems that process data at scale. But beyond the textbooks, it’s about cultivating curiosity. Take the time to experiment with projects outside your comfort zone, whether training a small neural network to recognize your cat or building a simple chatbot for your friends. AI isn’t a linear path—it’s a blend of discipline and creative exploration.”
Are there AI certifications to help experienced developers/engineers transition to AI?
“Certifications can help, but I’d suggest seeing them as tools, not tickets,” Clark says. “Programs like Google’s TensorFlow Developer Certification or AWS Certified Machine Learning Specialty can show you understand the tools, but how you use them matters. Think of it this way: owning a hammer doesn’t make you a carpenter; knowing when and how to use it does. For those looking to pivot, I’d recommend Andrew Ng’s Deep Learning Specialization on Coursera—it’s more than just lectures; it’s a mindset shift that helps you think about practical AI applications rather than abstract theories.”
“Google, Microsoft, and IBM offer certifications, and there are crash courses from places like MIT,” Sundar notes. “Organizations like the Singapore Computer Society also offer courses with certification after successful evaluation. If you're focused on learning, many open courses are available on platforms like YouTube and Udemy.
Many companies at the forefront of AI development offer certifications, but jobs themselves are not (yet) requiring AI certifications. Unlike platforms like Salesforce, certifications in AI aren’t currently viewed as proof you have a firm understanding of the technology or any particular model.
What programming languages are most important for a developer/engineer interested in AI?
Sundar says good ‘ol Python is a great start: “Python is a good choice because it has various libraries that simplify implementing core mathematical, probability, and statistics concepts. These libraries are also relatively efficient compared to other programming languages.
“From a job perspective, knowing an additional language like Java can be helpful, as many industry applications use hybrid solutions. Business functionalities may be written in different programming languages, so it’s beneficial to know more than one language to understand and work on both.”
Clark agrees: “For me, Python is the universal translator of AI. Its libraries—TensorFlow, PyTorch, and Scikit-Learn—simplify complex processes, letting you focus on the logic rather than the syntax. But if you want to dive deep into optimization and real-time applications, C++ becomes indispensable, especially when deploying models where milliseconds count.
“I also think Julia is a sleeper hit—its speed with numerical computations could be the next big thing for those focused on research-heavy roles. It’s about choosing the right tool for the problem, like a chef knowing when to use a paring knife versus a cleaver.”
What are some great open-source AI repos for developers looking to add to their portfolio?
Contributing to open-source repos is a great way to prove your knowledge and exhibit your skills. “When it comes to open-source AI, I believe quality over quantity is the key,” Clark says. “Hugging Face’s Transformers is a goldmine if you’re into natural language processing or NLP—it’s not just about using pre-trained models but understanding how to adapt them for your specific needs. fast.ai offers an incredible balance between deep learning theory and practical applications—you get to train models that are powerful yet simple to understand.
“For those wanting to show off data preprocessing chops, DVC (Data Version Control) helps track datasets and models like a pro, bridging the gap between machine learning and data engineering. It’s not just about contributing code; it’s about contributing to your understanding of AI’s ecosystem.”
Sundar advises involving yourself with repos for underlying frameworks, like TensorFlow, rather than chasing down great AI repositories. Displaying proficiency in the platforms and frameworks supporting AI may be uniquely valuable in these early days of AI.
