Pau Sabrià: the founder who keeps building from scratch
Pau Sabrià is the kind of founder who doesn't stop. He grew up in Barcelona — half Spanish, half Argentine — the son of an Olympic water polo player and a painter, raised in a family where taking irrational-looking risks was considered normal and even wise. He studied Telecoms at UPC, spent three years at BCG Barcelona, did his MBA at Columbia, and then did what most consultants only talk about: he walked away from a guaranteed career, took a loan against his parents' house, and built a company from a desk in Chinatown, New York.
That company was Olapic. Seven years after founding it with two co-founders, he sold it to Monotype for $130 million.
Then he started again.
Today Pau is the co-founder of Remotely Works — a talent marketplace connecting senior software engineers across Latin America with US software companies. The business has reached $40M+ in GMV, is growing at 40–45% per year, has 500+ engineers placed across 100+ client companies, and has never spent a dollar of its original $1.5M seed round.
In a conversation with Ramón Rodrigáñez, CEO of Nova, Pau unpacks both stories — and makes a sharp-eyed case for why AI is not eliminating senior engineers. It's making them more valuable.
What is Remotely Works and how does the business model work?
Remotely Works is a network of pre-vetted senior software developers in Latin America, placed into US software companies as extended team members. It operates as staff augmentation: the developer's contract sits with Remotely Works; the client gets the talent, the timezone alignment, and the operational simplicity.
The pricing model is deliberately different from the industry norm. Most IT outsourcing firms operate with opaque markup structures — you pay a blended hourly rate and have no idea what percentage reaches the engineer. Remotely Works works on a flat monthly fee per developer (which functions as a take rate of roughly 15–17%) on top of whatever the client and developer negotiate directly. The developer sees and agrees to their salary. There are no hidden markups.
The numbers:
MetricFigureGMV billed to clients~$40M/yearNet revenue (take rate)~$7M/year (~15–17%)Developers placed520+Client companies~100Annual growth rate40–45%Headcount (internal team)~40 peopleAverage developer salary$8,000–9,000/month
The average client has five engineers placed through Remotely Works. Senior developers on the platform can earn up to $15,000–20,000 per month — competitive with US salaries outside the top tech tier, and often equal to or better than what a B2B SaaS company in Alabama could offer a local hire.
"We make the process completely transparent. The client negotiates the salary directly with the developer. What the developer agrees to is what the developer earns. We charge a flat fee above that — and it never changes based on the salary, so I have no incentive to inflate it."
— Pau Sabrià, Co-Founder · Remotely Works
How Remotely Works finds and vets 500+ senior engineers — starting from open source
The supply side of any talent marketplace is the hard part. Remotely Works solved it with an unusual approach: instead of posting job ads and waiting for applications, they built a proprietary database of senior developers by mining open source contribution data across GitHub.
Here's how it works:
- Map all public open source repositories — every commit, every pull request, every contributor, indexed by language, library, and repository popularity.
- Build a page-rank-style quality score — a pull request accepted into a popular, high-quality repository is worth more than one accepted into a private or low-traffic project. Popularity of the repository acts as a quality signal, like an academic citation count.
- Estimate geolocation — contribution timezone patterns, language of commit messages, and other signals allow rough Latin America filtering.
- Cross-reference against LinkedIn profiles — connecting anonymous GitHub handles to real people with professional histories.
- Run a human screening interview — every candidate who enters the active network is interviewed by a Remotely Works recruiter in person (not AI-screened). The interview creates a video clip attached to the enriched profile.
Starting from 350,000 identified profiles, Remotely Works has built an active network of 7,000+ pre-vetted developers with profiles that go far beyond what LinkedIn offers — technical depth, code quality signals, video clips showing communication style and seniority.
The system was originally built using technology from a Spanish startup called Sourst (whose team Remotely Works hired after the company shut down). It is now further enhanced with AI — though the core insight, Pau notes, was never about AI. It was about using code quality as a proxy for developer quality, without ever asking the developer to take a test.
"The best developers don't want to take coding challenges. They're inundated with them. We went around the problem entirely — we let their actual work speak for itself."
How does Remotely Works protect against being cut out of the relationship?
Disintermediation is the structural risk of every marketplace. Once a client has a great developer, why keep paying the 15%?
Pau's answer is unsentimental: developer tenure is short. The average engineer placed through Remotely Works stays with a client for 18–24 months before moving on. This creates a constant need to re-hire — and clients who have tried to do this themselves typically come back. Building a dedicated Latin America recruiting operation requires infrastructure, a network, and months of time. For companies whose core competency is software, not recruiting, it rarely makes sense.
The second protection is the platform itself: 7,000+ pre-vetted profiles with enriched data and video clips represent years of investment that a client cannot replicate with a LinkedIn subscription and a recruiter.
The third is reputation and referral. The majority of Remotely Works' clients come from existing client referrals. A company that tries to cut Remotely Works out risks losing access to the network — and word travels in tight tech communities.
How Olapic went from a $50K idea to a $130M exit in six years
The founding story
Olapic started at Columbia Business School in 2010. Pau and his co-founders — José de Cabo and Luis Sanz — began with $50,000 raised from friends and family and a simple idea: collaborative online photo albums. The first commercial use case was weddings (one clear payer, multiple contributors). That attracted a corporate client — the New York Daily News — who wanted audience-contributed photo galleries. That led to a second pivot: user-generated content for e-commerce.
The insight was clean: when a product page shows real photos of real people using a product in real life, conversion rates go up. Olapic built the infrastructure for brands to collect, license, and display user-generated content at scale. IKEA, apparel brands, retailers — they all wanted to turn Instagram photos of their products into conversion tools.
The funding journey
RoundAmountMilestonePre-seed$50,000Columbia, co-founders onlyFirst raise$825,000Post-incubator, ~18 months after launchSeries A$5M~2 years in, ~$2M ARRSeries B$15MAfter e-commerce pivotAt exit~$13M ARRSale to Monotype at $130M
The acquisition
In early 2016, Monotype — a Nasdaq-listed company with roughly $200M in annual revenue, best known for owning the world's largest font library — approached Olapic about an acquisition. Conversations began at Easter. By July, the deal was closed at $130 million, plus retention packages.
Pau and his team stayed for three more years. The post-acquisition period was educational in ways the startup years weren't: investor calls, quarterly earnings guidance, working with a Chief Revenue Officer from Oracle, and dealing with an activist investor (the same fund that targeted Yahoo) who pushed Monotype to cut costs — including eventually the Olapic team.
The lesson Pau draws is not about the sector or the acquirer. It's about the deal structure itself: get clean proceeds at exit, because staying on as an employee after giving up control is a fragile position.
Is AI killing junior software engineers? What Remotely Works is actually seeing
This is the question Pau hears constantly — and his answer is more nuanced than most.
What has actually changed
Two things have shifted in quick succession. First, the LLM era: AI as a dramatically better question-answering and code-generation assistant. Second, and more recently, the agentic era: systems like Claude Code and OpenAI's Codex operating in the IDE or command line, executing multi-step engineering tasks rather than just answering questions.
The combination means that a skilled senior engineer today can produce significantly more than they could 18 months ago. The ROI on hiring a senior engineer has gone up.
The seniority pyramid is inverting
The traditional engineering org chart looks like a pyramid: many juniors, fewer mids, fewer seniors. Remotely Works is seeing that structure invert among its clients.
- Open junior positions: declining
- Open senior positions: rising
- Some clients have stopped hiring juniors entirely
The reason: an AI agent can do what a junior engineer used to do (write boilerplate, implement well-defined tasks, move fast on clear specs). What it cannot do is what a senior engineer does — understand why something is being built, make architectural decisions, write good prompts that specify the problem precisely, and supervise the output critically.
"The tools make you extremely productive if you are good at what you ask them to do. A senior engineer knows exactly what to ask and exactly what good output looks like. That skill is becoming more valuable, not less."
The Dan Brown paradox
Pau uses an analogy to explain the risk of over-relying on AI: if you're not a historian and you read Dan Brown, you might think you're reading rigorous historical analysis. The books are compelling because they trace a plausible line between real history and invented fiction — and you can only spot the sleight of hand if you already know the subject.
AI-generated code works the same way. If you don't know what good engineering looks like, the output can look brilliant while being subtly or dangerously wrong. Domain expertise is the prerequisite for using these tools at their full potential.
What else is changing in engineering orgs
Beyond the seniority shift, Pau identifies two structural changes:
1. The squad composition is changing. Traditional squads were 1 product manager + 1 designer + 5–7 engineers. That ratio is shifting toward fewer engineers per squad and more demand for product managers — people who can direct AI agents rather than just write code themselves.
2. AI is permeating non-engineering functions. Sales teams are coding (sort of). Finance teams are building internal tools. IT departments are transforming from laptop procurement into AI rollout and governance functions. Every one of these functions is starting to request dedicated engineering support — creating new demand.
"I'm an optimist without hope": Pau's view on AI and the future of work
Asked directly whether he's optimistic about the future of employment, Pau reaches for a phrase from a Stanford philosophy professor: I'm an optimist without hope.
He's skeptical of the people who say AI will always be a companion and that humans will always be needed in their current roles. He thinks many jobs will disappear — including, eventually, knowledge work like the consulting he did at BCG. But he doesn't think that means there's no point fighting for good outcomes. It means the transition matters enormously.
His frame: the Industrial Revolution and agricultural mechanisation destroyed millions of jobs in the short term, and that destruction was brutal and real. But human needs didn't cap out — they expanded into new forms. Running clubs and specialty coffee shops didn't exist 10 years ago as mainstream industries. The category of "very human experience that people will pay for" will grow.
His worry is not about the long run. It's about whether governments and institutions can absorb the transition pain fast enough.
Frequently asked questions
What is Remotely Works?
Remotely Works is a staff augmentation marketplace that places senior software developers from Latin America into US software companies. Developers are pre-vetted through open source analysis and human interviews. The company operates transparently: clients negotiate salaries directly with developers, and Remotely Works charges a flat monthly fee.
How much do software developers earn through Remotely Works?
The average Remotely Works developer earns $8,000–9,000 per month ($96,000–108,000 annually). Senior developers on the platform can earn up to $15,000–20,000 per month. This represents roughly one third of the total compensation cost of hiring an equivalent engineer locally in the United States.
What is Olapic and how much was it sold for?
Olapic was a user-generated content platform that helped e-commerce brands collect, license, and display real customer photos on product pages to increase conversion rates. It was founded in 2010 by Pau Sabrià, José de Cabo, and Luis Sanz at Columbia Business School. The company was acquired by Monotype in 2016 for $130 million, at approximately $13M in ARR.
Is AI replacing software engineers?
Not at the senior level — the opposite appears to be happening. Remotely Works data shows that open junior positions are declining while senior positions are rising. Some clients have stopped hiring juniors entirely, replacing that function with AI agents supervised by senior engineers. Senior developers with strong architectural judgment are more valuable now because they can direct AI tools effectively; junior developers without that foundation benefit less.
How does staff augmentation work for US companies hiring in Latin America?
Staff augmentation means a third-party company (like Remotely Works) handles the employment, compliance, and payroll of developers who work full-time within a client's team. The client company benefits from Latin America's lower cost base, overlapping time zones with US east coast hours, and strong engineering talent — without setting up a foreign subsidiary or managing international HR.
How did Pau Sabrià fund Remotely Works?
Remotely Works raised $1.5M at its founding in early 2020 — a round that was closed in a single weekend via WhatsApp messages from investors who had known Pau from the Olapic days. The company has never raised a subsequent round. As of 2026, the original $1.5M remains largely unspent, and the business is profitable with strong cash generation.
Last updated: April 15, 2026
🔑 Key takeaways
1
Transparent pricing is a real competitive advantage in opaque markets. The IT staffing industry runs on hidden markups. Remotely Works' flat-fee, salary-transparent model is a genuine differentiator — and attracts better developers because they know exactly what they'll earn.
2
Open source is an underused talent signal. A developer's GitHub contribution history tells you more about their quality than any coding test. The best developers won't take tests — but they've already shown their work in public.
3
AI is inverting the engineering pyramid. Junior positions are declining; senior positions are rising. The teams that move fastest are those with fewer, more experienced engineers directing AI agents — not those adding junior headcount.
4
Profitable bootstrapping is a legitimate path. Remotely Works is at $40M GMV, growing 40%+ per year, and has spent almost none of its seed capital. The business generates cash, gives founders full control, and can compound indefinitely. Not every business needs venture scale.
5
The best time to raise is when you don't need it. Remotely Works raised $1.5M in a weekend via WhatsApp — because investors already knew the team from Olapic. Reputation compounds. Solve problems before you have cash, and the cash will come easily.
6
Domain expertise is the moat that AI can't cross. AI tools produce output that looks competent to a non-expert. The engineer, the marketer, the consultant who actually knows the field can spot the errors — and direct the tool to produce something genuinely good.
PS
Pau Sabrià
Co-Founder · Remotely Works & Olapic
Telecom engineer (UPC Barcelona). Consultant at BCG Barcelona. MBA at Columbia Business School. Co-founded Olapic in 2010 — a UGC platform for e-commerce brands — which was acquired by Monotype in 2016 for $130M. Co-founded Remotely Works in 2020: a staff augmentation marketplace connecting senior Latin American software engineers with US companies, now at $40M+ GMV growing 40% annually. Nova community member.
Nova: where ambitious professionals find each other
25,000+ vetted professionals across 80+ countries — including founders like Pau who've built, sold, and built again. The network where your next co-founder, investor, or mentor is one introduction away.
Membership is by nomination.
Apply to Nova →
Further reading
- Remotely Works — the platform Pau built for placing senior Latin American engineers into US teams.
- All Nova Podcast Pills — the full archive of founder and operator stories from the Nova community.
- Fernando Usera: the company that was stolen, and the organic grocery chain built in its shadow — another Nova founder story about building, losing, and rebuilding from scratch.
