Yang Li discusses how the role of the software engineer has been remoulded by increased automation and innovation.
Software engineering first emerged as a viable career route for skilled professionals in the 1960s and since then has developed into a discipline that is arguably defined by its rapid evolution. In the 1970s the main focus was on creating smarter, less costly software. Through the 1980s and 1990s the industry was altered once again by the introduction of object oriented programming.
Fast forward to today, where advanced languages, cloud computing and AI have rendered the role of the software engineer virtually unrecognisable when compared to its early beginnings. For Yang Li, the co-founder and COO of software development company Cosine, it is this consistently forward motion that makes technology such an exciting and dynamic area to work in.
“Unlike other STEM fields, like biotech, that might be more resource-constrained or slower-moving, technology allows you to iterate rapidly and continuously improve solutions,” he told SiliconRepublic.com. “That combination of speed and tangible impact is what keeps me passionate about this field.”
Now, as we stand at the start of a new year, for Li the role of the software engineer is morphing once more, from that of a code writer to an AI orchestrator, a professional who can coordinate and manage the deployment, integration and interaction of various AI components within a system. But with this comes some challenges.
Exciting times
“A software engineer’s main challenge today is not writing code, it’s figuring out architecture, planning and reasoning,” he explained. Separately to that, he noted that gone are the days in which companies will mass hire engineers and rather than scaling though headcount, organisations are looking to grow and evolve via computational power.
“Both of these trends mean engineers need to think bigger. They need to aim for skills that allow them to architect solutions across the full stack and direct AI to build them. We’re already seeing this with our own team, our top engineers are working on multiple tasks simultaneously, orchestrating AI to handle implementation while they focus on outcomes.”
For Li, issues also creep in with code reviews as, while AI can generate code far quicker than before, even autonomously, the review process still demands significant human oversight for large quantities of information.
Engineers can become too reliant on this AI-generated code and fail to properly understand the underlying systems, which makes the debugging process far more difficult. “You end up with engineers who can produce solutions but can’t explain their logic or decisions.”
According to Li, despite the challenges there are a number of concrete examples that speak in favour of the introduction of advanced AI into software engineering careers, namely, how it can expedite tasks that would typically take hours, which now take mere minutes.
“When decisions are automated, it becomes harder to figure out how and why certain decisions were made. That said, it’s still easier to review and critique code than write it from scratch. Even if you need to troubleshoot something, it only takes about 50pc of the effort compared to doing everything manually yourself.”
Trust the shifting sands
We have seen it in the past with social media, digital literacy and AI, where at a certain point educational institutions and companies have to further develop their curriculums and training in order to give students and professionals the opportunity to grow their skillsets and embrace new career routes.
“Software development is evolving from a trade into a creative profession. Those who can orchestrate AI systems and think strategically about product development will be in high demand. Just as ATMs freed bank employees to focus on higher-value services, AI development tools will enable engineers to tackle more ambitious projects while automated systems handle routine optimisation and maintenance.”
But for Li, how organisations and institutions approach AI skills development in software engineers will be key to its overall effectiveness. He is of the opinion that, once given the tools and the foundational basics, developers should be encouraged to be creative, build their own processes and to strategise.
“It’s like trying to enforce training on the exact order someone should solve a coding problem, different people will be more creative and have different approaches. While you can teach people how to use the tools, you can’t standardise how they’ll leverage them creatively. The key is giving access and letting developers find their own ways to be effective with these tools.”
Ultimately, Li predicts a seismic shift in the software industry, where companies will depend on an elite group of engineers with an in-depth understanding of advanced AI tools.
“Most software development teams as we know them today will change. Instead of having large teams of hundreds of developers, we’ll see small groups of 10 to 20 people using AI to do the same amount of work. These teams will focus on guiding and checking AI’s work rather than writing code themselves.”
In the end, when it comes to hiring, assessments, workplace seniority, HR matters and further developing skills, the traditional methods may no longer be functional or relevant and organisations may “need to rethink the entire economic model of software engineering careers” altogether.
Don’t miss out on the knowledge you need to succeed. Sign up for the Daily Brief, Silicon Republic’s digest of need-to-know sci-tech news.
