The technology industry has always had two native languages: the language of how things should feel, and the language of how things get built. People fluent in both are rare. The products they make are unmistakable.
Maitrik is one of those people. An engineer by degree and a designer by nature, he made an unusual choice early in his career: he would not pick a side. He would become the translation itself, the place where vision becomes a system without losing what matters.
He spent his first decade translating between design and engineering, in the field that came to be called DesignOps. He has spent his second translating between research science and engineering, in the field now known as MLOps. The industry invented a title for this kind of person along the way: design technologist. Maitrik was practicing the discipline before it had a name.
The First Decade: Closing the Gap
The conviction seemed almost contrarian at the time: design quality was an engineering problem. The gap between a beautiful mockup and a shipped product was not a talent gap but an infrastructure gap. Over the next decade at Walt Disney, Sony, and DocuSign, Maitrik built his career on proving it.
At Sony, that conviction took the form of a design framework deployed across 110 countries and 30 languages, a system that had to survive localization, right-to-left scripts, and cultural variation in color and typography without fracturing into inconsistency.
At DocuSign, it became the Design System that served as the backbone of DocuSign Agreement Cloud surfaces through the company’s public offering, supporting a product organization operating at over a billion dollars in annual revenue. The system meant that a button in the mobile app behaved identically to a button in the enterprise dashboard not because a meeting decided it, but because the infrastructure guaranteed it.
The discipline he was helping to define eventually earned its name. Its core argument was Maitrik’s argument – that the bottleneck in great design was never creativity. It was orchestration.
The Second Decade: Frontier
Then came the opportunity that would test whether the translator’s skill set could survive a change of language entirely.
When Maitrik joined Apple’s AI organization, the language on one side was no longer design. It was research science – people whose intuitions lived in probability distributions and loss functions, not color systems and type scales, but who cared just as obsessively about getting it right. The other side was still engineering. And the translation between them carried higher stakes than anything Maitrik had encountered in design.
He recognized the shape of the problem immediately. He applied the same rulebook.
Every model is a mirror of the process that trained it. The data labels, the tools, the quality controls are not logistics. They are the model’s DNA. Across the industry, this infrastructure has historically grown organically – tools built by different teams, operating under different standards, with training cycles measured in months. Scaling it is one of the hardest unsolved problems in applied machine learning.
Maitrik’s work brought the design system playbook to that problem: unified platforms, shared standards, infrastructure that teams could trust without thinking about. Annotation cycles that once took months were compressed to a fraction of the time. The labeled dataset available for model training expanded by orders of magnitude. The models improved. The infrastructure disappeared from view. The discipline had found its second name by then: MLOps.
“Everyone wants to work on the thing,” Maitrik has said. “The best teams are only as good as the infrastructure beneath them. A designer needs components they can trust. A model needs labels they can trust. That’s always been my territory.”
Built to Last
Today, Maitrik leads platform initiatives at Apple: developer experience tooling for the company’s GenAI platform, and an ML orchestration service built on industry-standard workflow infrastructure. His team builds the systems that other engineers depend on to ship AI features at consumer scale, reaching hundreds of millions of people who will never know his work exists.
That invisibility is the throughline.
Maitrik has come to believe the two disciplines he bridged are, at their core, the same challenge: how do you scale human judgment? A design system scales the judgment of a designer across thousands of product decisions. An annotation platform scales the judgment of a human expert across millions of training examples. Different specifics. Identical shape.
“The field that figures out how to scale judgment reliably will define what AI looks like in ten years,” Maitrik has said. “That is not a model problem. It is an infrastructure problem. It always has been.”
Two decades ago, Maitrik chose to live where the languages meet. It turns out that is where the future gets translated into something real, and he has been doing the translating the whole time.
