An office has a useful trick. It makes work visible, even when very little work is happening. A person at a desk looks employed. A person somewhere else requires an explanation. Milena Berry had already spent years testing the limits of that arrangement when she co-founded PowerToFly in 2014. She had led technology for an international campaigning organization, helped support millions of members and raised a family. Her colleagues were scattered around the world. The missing office had apparently failed to ruin everything.
What interested her was whether other people could have the same arrangement. A woman with the skills to build software might also have a child to collect, a home far from a technology hub or a life she had no intention of packing into a moving van. Employers could call this a shortage of candidates. Berry had reason to suspect a shortage of imagination.
Her response was a company that matched women with remote jobs. Twelve years later, PowerToFly is applying its network to artificial intelligence. The machinery has changed. The question of whose expertise gets noticed has proved remarkably durable.
First, make something peculiar
Berry was born in Bulgaria and moved to New York in 2001, arriving a week before September 11 to study at New York University. She entered the Interactive Telecommunications Program at Tisch, completing her studies in 2003. Robotics, wearable technology and social software were among her subjects. These were ways of asking what people might do with computers once computers escaped the conventional computer-shaped box.
Her own account of those student years is pleasingly untidy. She programmed microchips, invented musical instruments and staged interactive audiovisual performances. She toured Europe and North America with an invention called the VoiceMutator. Through LEMUR, she participated in robotic musical installations in New York galleries. Before becoming a CEO concerned with the organization of work, she had been experimenting with the organization of sound.
The connection is tempting, though it should remain a connection rather than an origin myth. A wearable device and a hiring platform are different things. Still, both invite a practical question: must a familiar activity keep its familiar form? Berry’s early work left room for answers stranger than a better keyboard.
Berry’s graduate experiments included wearable technology, musical instruments and interactive installations.
Twenty colleagues, six continents
At Avaaz, the online campaigning organization, Berry became chief technology officer. The technology operation she helped develop supported growth to 33 million members. Her technical team numbered about 20 and worked across six continents. For someone later expected to persuade employers that remote work could function, this was useful preparation. She could point to an organization that already depended on it.
She also knew the hiring problem from the other side of the desk. Engineering managers, she explained, would secure a budget, face pressure to fill a role quickly and discover that their applicant pools were overwhelmingly male. The urgency encouraged familiar recruiting habits. Those habits helped reproduce the same pool. Wanting different results did surprisingly little to change the process that produced them.
Her advice was to examine the requirements themselves. Did the role truly need a particular university pedigree? Was a formal computer science degree essential? Each condition could be defended in isolation. Taken together, they might quietly exclude people who could do the job. Recruiting had become a series of decisions that looked neutral until someone inspected who was left outside.
Two founders, no office lease
Katharine Zaleski brought a different professional history to the partnership. She had worked in digital journalism, including at The Huffington Post and The Washington Post. After having her first child, she was reconsidering how a demanding career could fit around family life. Berry approached her with the idea of connecting women to technical work they could do from home. In August 2014, PowerToFly launched.
The early pitch was concrete. Find skilled women, vet their work and connect them with employers willing to hire across distance. A developer in Romania could be relevant to a company in New York without first becoming a New Yorker. Location, previously treated as a qualification, could become a detail to manage.
The founders also had to practice what they sold. By July 2015, they and their 35 employees worked remotely, with no office space. Nearly 1,000 companies had posted jobs on the platform, including Hearst, BuzzFeed and The Washington Post. The listings extended into marketing, sales and customer service. The experiment was becoming less narrowly technical.
That June, PowerToFly announced a $6.5 million Series A led by Crosslink Capital, with Hearst Ventures and Lerer Hippeau participating. Together with its earlier seed funding, the company had raised $7.5 million. Money brought its own practical questions: what to build, whom to hire and how to turn an argument about work into a business that paid its bills.
Getting the job was only the beginning
A job offer can solve one problem while exposing another. Berry later described candidates arriving at apparently desirable companies and discovering that they did not feel they belonged. Better recruiting could get someone through a door. The daily experience on the other side required its own attention.
In a discussion about inclusion, she recounted a company whose board tied its chief executive’s bonus to diversity goals. The intention sounded serious. The trouble was that the incentives did not continue down through management. The people whose decisions would produce the result had received an announcement without a corresponding change to their own measures of success.
Berry’s prescription was specific: give the work a budget and make responsibility travel through the organization. Appointing someone to oversee diversity without resources did little to change what the company could accomplish. This was the operator in her talking. A good intention cannot authorize its own purchase order.
PowerToFly’s remit broadened accordingly. Recruitment sat alongside retention, learning and the culture people encountered at work. Berry also described how thinking about her own overlapping identities as a woman, caregiver, queer person and immigrant helped widen the company’s focus beyond gender. A person could encounter several barriers at once. A business built around access would have to notice more than one.
“Push yourself to the other side of fear.”
Milena Berry, on leadership, 2016
A little less mythology about the CEO
For all the public conversation about purpose, Berry is willing to discuss the unglamorous requirements of running a company. In 2023, she warned prospective CEOs that they needed to understand their numbers: profit and loss, unit economics and the costs behind delivery. She admitted she could procrastinate over the figures herself. The spreadsheet, it seems, does not become more charming simply because the mission is worthwhile.
Her advice about leadership also makes room for gaps. A chief executive needs colleagues with expertise she lacks, she has said, and must learn to make decisions while afraid. Neither thought offers much material for a bronze statue. Both are useful when people are waiting for an answer.
Her family remains part of the account rather than an inconvenient footnote. She has four children, and a 2025 conversation about her immigrant experience included raising them with her wife. The argument for flexible work has always had a domestic scale as well as a commercial one. A career occupies a life. It does not acquire the whole property.
The missing skill gets a place in the plan
By September 2023, Berry was describing the next problem: connecting qualified people to employers was insufficient when someone needed to learn a skill first. PowerToFly could help build the qualification as well as arrange the introduction. In May 2024, the company acquired Skillcrush, the education business founded by Adda Birnir.
Skillcrush brought flexible, mentor-led training in fields including coding, design and AI. At the announcement, it had trained more than 20,000 students; 80 percent identified as women and more than 49 percent as BIPOC. These figures described Skillcrush’s students, rather than PowerToFly’s own community. The acquisition joined an education operation to a company already working on hiring.
For Berry, the attraction was a route from learning to a conversation with an employer. That route matters because completing a course and getting a job are separate achievements. The acquisition gave PowerToFly more of the intervening work to do. It also made an earlier client relationship rather interesting in retrospect: Skillcrush had appeared among PowerToFly’s early customers a decade before.
- 2014Connect women with remote jobs.
- 2024Add technical training through Skillcrush.
- 2026Bring specialist judgment to AI work.
Now the machine needs a second opinion
In September 2026, PowerToFly announced an October 14 launch as an AI workforce company. Its services include recruiting specialists and supplying people for model training, evaluation, red teaming and quality assurance. These jobs depend on people who can recognize when an answer sounds convincing and is still wrong.
PowerToFly now describes a community of more than 1.1 million professionals across 190 countries. Within its network, it reports more than 380,000 domain and functional experts and 65,000 AI specialists. Those are company-reported counts of different groups. They give a sense of the operation’s reach without telling us how many people have been hired or how any individual project performed.
The personal continuity is easier to see. At Avaaz, Berry had worked with expertise distributed across continents. At PowerToFly, she helped employers look beyond their usual recruiting radius. Now she is asking them to recognize the judgment needed to assess a machine’s work. A specialist’s years in a field are relevant precisely because a fluent answer cannot certify itself.
For Berry, this leaves a familiar practical task. Find the person who understands the work. Give that person a route into the project. Build an arrangement in which the contribution can actually be used. The office may still be there, with its desks and its reassuring evidence of attendance. Her career keeps returning to the people it was too easy to overlook while counting the occupied chairs.