Utkarsh Amitabh says he definitely wasn’t looking for a new job in January 2025, when data labeling startup micro1 was approached to join its network of human experts who help companies train artificial intelligence models.
The 34-year-old UK-based entrepreneur already had a busy schedule as an author, university lecturer, founder and CEO of Network Capital, a global mentorship and career platform, and a PhD student. At Oxford University’s Said Business School. There was also a newborn baby at home, he told CNBC Make It.
Ultimately, Amitabh agreed to take on the additional role, admitting that “intellectual curiosity attracted me.” He added that the prospect of training enterprise AI models felt like a good fit with his own background in “business strategy, financial modeling, and technology.”
In fact, micro1 says it employs experts with deep knowledge across a wide range of specialties, from doctors and lawyers to engineers. Amitabh, a self-described “deep generalist,” seems to fit the bill.
He has a bachelor’s degree in mechanical engineering and a master’s degree in moral philosophy, and has spent over six years working in business development at Microsoft in roles focused on cloud computing and AI partnerships. His previous publications include a book on the “side hustle revolution” and a master’s thesis on how AI will impact the nature of performance.
The opportunity at micro1 seemed like a “natural” fit, Amitabh says. He also appreciated the flexibility of part-time freelance roles. He works an average of about 3.5 hours each night, usually after his 1-year-old daughter goes to bed, he says.
“It didn’t seem like an add-on, but something I could use to further my interests for a limited time a week,” Amitabh says.
Amitabh currently earns $200 an hour for his job training AI models for micro1, based on pay stubs seen by CNBC Make It, and a company spokesperson confirmed that Amitabh has been making nearly $300,000 for his job since January, including project completion bonuses.
At the same time, Amitabh says that “money was not a motivator” as the role overlapped with his personal and professional interests, especially considering he was already earning a living from other jobs. Still, he added that he believed “fair pay is a core value” and felt the compensation was “respectable” for work that required a high degree of expertise.
“You have to pay close attention to detail.”
Founded in 2022, micro1 has built a network of more than 2 million experts who train AI models for clients such as large AI labs including Microsoft and Fortune 100 companies that develop their own large-scale language models for their own employees, according to TechCrunch. Micro1 was recently valued at $500 million, and its competitors include big startups like Mercor and ScaleAI.
Daniel Warner, chief marketing officer at micro1, said in a statement that a network of experts like Amitabh forms the “backbone of our data quality,” adding, “Today’s AI models have already absorbed most of the publicly available knowledge and are making real progress.” “We are driven by subject matter experts who can challenge, refine, and effectively think through models. Thanks to ‘human data’ generated by true experts, we are able to deliver best-in-class results to leading AI labs and Fortune 100 companies.”
Training an AI model involves feeding large amounts of information and scenarios into algorithms to form large datasets. The model is then refined over time by testing it with prompts that answer questions or suggest solutions to problems. For example, ask an AI agent to track expenses, grow a project, or create a new budget for a business unit within your company.
Amitabh says that many of the projects he works on are confidential and involve “looking at complex business problems that a typical user, business owner, or business owner might have and breaking them down into smaller parts.”
Similar to prompt engineering, this part of the job requires breaking down each problem into clear, concrete language that “machines can understand” to ensure the model can return accurate and appropriate responses, he added.
When a model’s response has errors or deviates significantly from the parameters of the original question or problem, Amitabh identifies where the “point is missed or subtlety is lost” and takes action so that the model’s dataset can be adjusted to improve and then tested again. It’s a trial-and-error process, he says, and can take “several hours” per problem set.
“You have to pay close attention to detail and be aware frequently of the mistakes that humans and machines can make. The process of immersing yourself in it allows you to discover more about the types of mistakes that exist,” says Amitabh.
The job is “intellectually very demanding,” he says, especially as AI models are constantly learning and improving, requiring even experts like Amitabh to level up their knowledge base and creative thinking skills.
“The ultimate goal is actually very energizing,” he added. “Machines and humans, the way we approach this will tell us whether we can level up the output of the problem you asked and other types of problems that may be related to it.”
AI and jobs: “The trillion dollar problem”
Amid the rise of AI in the workplace, a concern for employees in most industries is whether advances in technology will ultimately eliminate the need for human workers, or at least significantly change their roles. So, is Amitabh worried that leveraging his expertise to train AI models now could reduce career opportunities for himself and others with similar backgrounds in the future?
“This is a trillion-dollar question,” he said, noting that people typically fall into the “techno-optimist or techno-pessimist” camp when it comes to how they view the impending AI revolution and its impact on the labor market. “I like to think of myself as somewhere between a techno-optimist and a techno-realist,” he added.
Amitabh acknowledged that there will definitely be “growing pains” as more companies implement AI tools into the daily activities of their employees, which will likely result in a significant number of job cuts, an impact that HR leaders say is already starting to take place.
But he’s also in the optimistic camp that expects AI to eventually create more jobs to make up for these losses. For example, the World Economic Forum’s January 2025 analysis predicts that AI will be a disruptive but ultimately beneficial force in the global labor market, adding nearly 80 million net jobs by 2030.
Ultimately, Amitabh says he takes a more philosophical view. He believes that the knowledge of both humans and machines is not a “finite” resource, that humans and machines will always have a symbiotic relationship, and that their progress requires permanent cooperation.
“It’s also possible that this fear of AI, collectively, allows us to learn better, improve our own skills, and frame questions about ourselves differently,” he said, adding, “So I’m not worried about[the idea of]AI Doom at all, because I think it’s doing a lot more good than bad.”
Want to stand out, expand your network, and land more job opportunities? Sign up today for CNBC Make It’s online course, “How to Build a Standout Personal Brand: Smarter Online, In-Person, and at Work.” Learn how to showcase your skills, build a great reputation, and create a digital presence that no AI can match.
