Pakistan's IT exports vs one Tata subsidiary

From Lahore to a San Francisco garage

The episode opens with Muzamil framing the conversation around a single economic argument: Pakistan’s most urgent need is to shift toward export-oriented industries that bring dollars in rather than push them out. Of all the options available, he says, IT services is the lowest-hanging fruit. That framing sets up the guest — Qasim Asad Salam, co-founder and CEO of Remotebase — as a practitioner of exactly that thesis.

Qasim was born and raised in Lahore, educated at Beaconhouse Defence and then LUMS, where he studied economics and political science rather than computer science. The reason was bureaucratic: Pakistani universities require physics and chemistry for a CS degree, and Qasim hadn’t taken chemistry. He had, however, been building websites since he was eleven. By the time he reached LUMS, he was already freelancing and saw little utility in a degree whose selling point — a 40,000 to 50,000 rupee starting salary — he was already exceeding on Upwork.

What LUMS did give him was societies. “Societies operated like full businesses,” he explains. As a freshman, Qasim became director of IT and creativity in seven different societies simultaneously, running events, managing budgets, and leading teams of two to three hundred people. It was, in effect, his first operator training.

Campus Feed and the angel investor who opened the door

While still at LUMS, Qasim built Campus Feed — an anonymous social network for university students where taboo topics like religion and politics could be discussed freely. Within two weeks, 80% of the LUMS campus was on the platform. Daily engagement exceeded Facebook’s at the time, running above 25 minutes per user per day. The app eventually scaled to ten universities and over 20,000 daily active users.

Campus Feed never made money. But it made something more useful at that stage: it got Qasim to Silicon Valley. An angel investor from the Valley attended a startup event where he pitched, paid for his visa and flights, and put him in a house with 18 other founders — half Indian, half Pakistani. “As long as I can get you from Pakistan to The US and from the airport to my house and back to Pakistan, I have done my job,” she told them. The rest was pressure-cooker learning.

Qasim describes the cross-border friendships that formed in that house as one of the more striking moments of the conversation: “Nations fight against nations but the connection at a personal level — there is no animosity there.” Those friendships, he says, are still active today when he visits Silicon Valley.

Campus Feed also built his O-1 visa case. The extraordinary ability visa requires meeting three of eight criteria. Media coverage from the app, his leadership role in an organization, and his mentorship work at Plan 9 filled six or seven of the eight criteria before he even applied.

Plan 9, Plan X, and the limits of Pakistan’s early startup ecosystem

Muzamil asks Qasim about his time at Plan 9, the government-backed incubator in Lahore. Qasim is measured in his assessment. Plan 9 gave him a place to work, 100,000 rupees a month split across five people, and trips abroad. He slept there on some nights. He is grateful for it.

But he is also honest about its ceiling. “Your mentor should be somebody that’s two steps ahead of you or one step ahead of you,” he says. In 2016 and 2017, Pakistan’s startup ecosystem was too young to produce those mentors. The people who came in to advise were often at the same stage as the founders they were advising. Pataari, the most advanced startup in the cohort at the time, had not yet raised two or three million dollars. Nobody in the entire Plan 9 program had.

Later in the discussion, Qasim notes that the quality of startups entering Plan 9 has actually declined since his time — not because the program got worse, but because better alternatives now exist. NIC at LUMS, improved access to capital through 2021, and better external mentors have drawn stronger founders elsewhere.

How a software house became Remotebase with 45 days of runway

After Plan 9 and Plan X, Qasim moved to the US on a tourist visa, registered a company, converted to an O-1, and spent eight and a half months visiting cities with no fixed base, distributing flyers at co-working spaces to find clients for his software house, CreativeMorph.

Then COVID hit. Two of their largest clients disappeared. The startup they were running alongside the software house — a performance tracking tool for engineers — was not gaining traction. “We were down to about thirty, forty-five days of money left,” Qasim says. He was sleeping in a garage in San Francisco, eating three-dollar meals from Trader Joe’s, and burning everything he earned.

He came back to Pakistan and started writing to his US network. Friends needed engineers. He placed them manually. Then he and his co-founder decided to stop operating like a software house and start building a platform. Every manual step in the matching process became a target for automation. That platform became Remotebase.

The platform model: 12,000 applicants, 8% make it through

Muzamil asks Qasim to explain how Remotebase actually works — both for engineers looking for work and for companies looking to hire.

The model is a bench-based staffing platform. Engineers apply, complete technical assessments calibrated to their experience level, and go through video interviews — all on the platform. Ninety-two to ninety-three percent are screened out automatically. The remaining eight percent are interviewed by engineers already working at Remotebase. Those who pass are hired onto an internal bench. When there is no client work, bench engineers build Remotebase’s own internal tools. When a client needs someone, the match is made.

“Last month, over 12,000 people applied,” Qasim tells Muzamil. “A software house in an entire year might not see that many applicants.” The model is explicitly designed to avoid the inventory mismatch problem Muzamil identifies — where an engineer goes through the full process but there is no immediate opportunity. The bench absorbs that gap.

Clients get longer-term engagements at a lower price point than competitors like Toptal or Andela, which focus on short-term, high-premium placements. Remotebase’s earliest clients, if they have not gone out of business, are still with the company today.

The talent gap and why software houses are making it worse

Muzamil pushes on the hardest question in the model: Pakistan produces roughly 5,000 computer science graduates a year, and Qasim estimates that only 2,000 to 2,500 of those are genuinely technically capable. The entire market — Systems, IBM, Teradata, and everyone else — is competing for the same small pool. How do you go from 2,500 to 25,000?

Qasim’s answer has two parts. The first is structural: the gap between universities and the job market is enormous, and most software houses are making it worse. He describes a common pattern where a senior engineer with five or six years of experience takes on ten different clients simultaneously, puts junior engineers underneath him, and those juniors never interact with a client directly. Their exposure is zero. “The entire industry is running on jugaad,” he says, “but that is not the solution.”

The second part is the buddy system. Every Remotebase engineer has a clause in their contract requiring them to give one hour a day back to the company. That hour goes toward mentoring one or two junior engineers. “You are not growing linearly. You are growing exponentially,” Qasim explains. Five becomes a hundred, a hundred becomes three hundred, three hundred becomes nine hundred. The compounding is real — but it requires someone to absorb the cost of the unbillable training period, which is where government could theoretically help and, in Qasim’s experience, consistently fails to follow through beyond the press release.

He gives Muzamil a concrete proof point: one of Remotebase’s first hires started at 70,000 to 80,000 rupees as a fresh graduate from NUS. Two and a half years later, he is at Amazon earning over $200,000 annually. Another is at Tesla. The grooming works when the exposure is real.

Pakistan’s IT exports versus one Tata subsidiary

One of the sharpest moments in the conversation comes when Qasim puts Pakistan’s IT sector in global context. Pakistan’s total IT exports are approximately two to two and a half billion dollars. Tata Consultancy Services — a single subsidiary of a single Indian conglomerate — exports twenty-five billion dollars in IT services annually. That number, Qasim points out, exceeds Pakistan’s entire current account deficit.

Muzamil adds a related point about textile exports: the headline number of fifteen to sixteen billion dollars carries significant input costs — imported cotton, dyes, raw materials — that reduce the net figure considerably. IT services, by contrast, are almost entirely net. “Every dollar earned is a dollar in,” Muzamil says. Salaries stay in the country. Even energy costs, the largest non-salary expense, go toward productive consumption rather than imports.

“All it takes is one company,” Qasim says. The opportunity has not been lost. He sees 2020 to 2030 as a decade of innovation, driven partly by the wave of engineers laid off from large US tech companies who will now go out and build.

Generative AI: the gold rush that has already started

Qasim is direct about where he sees the next platform shift. Generative AI has already inverted the expected order of automation. The assumption was that repetitive, low-skill tasks would be replaced first and creative, high-skill work last. The opposite has happened. ChatGPT passes conversational tests. Stable Diffusion generates original art. “AI has already passed the Turing test at the highest tier,” Qasim says.

His advice to younger people is to treat generative AI the way early mobile developers treated the iPhone in 2007 — a five-to-ten-year window before the space becomes crowded. He gives two examples from his own network. A friend built a WhatsApp integration with the OpenAI API in eight to ten hours; it now has 20,000 to 25,000 users. Another company lets users photograph a room and generate interior design options from a text prompt. Jasper AI, which trained OpenAI’s model to write blog posts, raised $120 million and started as a weekend project.

“Get into this,” Qasim says. “This is the next transformational platform.” The skill required is not deep AI research. It is the ability to identify a narrow problem and use existing models to solve it.

By the end of the conversation, Muzamil asks Qasim where he sees Pakistan in 2050. Qasim is blunt: the political system will continue to deteriorate and he has zero hope for it. But he sees governments globally losing jurisdiction as remote work makes taxation and borders harder to enforce. Countries like Pakistan — sixth largest population in the world, one of the highest English-speaking populations, one of the youngest median ages — stand to benefit disproportionately from that shift. “I see a lot of hope that economic growth will be quite significant,” he says, “but absolutely zero hope from the political system.”

The bet, in other words, is on the people rising without waiting for the system to fix itself. Remotebase is, in a narrow sense, a test of whether that bet pays off.