In 2002, Wahid Choudhury was a Bangladeshi physicist with an unfinished PhD, four years of software experience at a London startup, and a problem his wife didn't exactly agree with: he wanted to move back to Dhaka.
He had left King's College London a few years earlier with an MPhil instead of a PhD, having concluded that "mediocre physicists who operate at the bottom rung actually don't have much prospect in this world," while "mediocre developers have many opportunities." He'd been doing computational solid-state physics, modeling what happens electronically when carbon atoms inside diamonds are displaced by iron or cobalt, but the programming side of the PhD interested him more than the physics itself.
What pulled him toward software was a feeling he kept coming back to: "I'm given a problem, I think about the problem, I program, I solve the problem, and I get a result. I feel happy with the closure, I go home. I have a good night's sleep." A sense of accomplishment. Physics rarely gave him that. A paper's open question could still be open twenty years later. The very problem he worked on for his PhD remains unsolved today, and he says it "doesn't take long to realize you are nothing in this game."
So while finishing his MPhil, he'd taken a part-time job at a London software startup, gone full-time the following year, and stayed for about four years. He watched the company grow from four or six people to a hundred within three months, after a Series A and then a Series B.
It was also a formative experience for Wahid, getting a front-row seat to watch how building and running an organization work. In particular, he mentions the fascinating ideas he absorbed from the company's CTO, a self-taught engineer with no formal education beyond A-levels who, in Wahid's words, became "so good at programming and computing that he held very important positions in some very large organizations." Wahid says plainly: "There are a lot of things that I copied from his style of doing things that I later applied at Kaz."
None of that was the plan. The plan, such as it was, didn't really exist. Wahid describes himself, even now, as someone who doesn't drive toward goals the way most founders are expected to. "Life to me kind of flows and I just flow with it,” he says. “You wouldn't find that spark in me if you want to compare me with typical founders."
The decision to leave London had nothing to do with career strategy. "My father and mother were alone and they were quite old, and my siblings live abroad. I felt that at least one of us should be with my parents," he says. "Another reason was I like Bangladesh. I have always liked Bangladesh." He adds, almost as an afterthought: "I wasn't thinking too much about the career, I just decided to jump." His wife, whom he'd married that same year, did not love the idea. There was "a bit of an argument," he says, much as there had been when he dropped the PhD. But she'd known him long enough to know what she'd signed up for.
Returning meant taking a normal job. A senior contact from his Dhaka University’s physics department, Sumon Ahmed Sabbir, ran BdCom, Bangladesh's third-oldest ISP, which was putting together a software wing. Wahid joined as a Senior Software Engineer, writing C++ to interface software with VoIP hardware switches, an unusually hands-on work for a developer, and work he found genuinely interesting.
But the pay in Bangladesh's software industry was relatively low, so in 2003, he started taking unpaid leave from BdCom to take outside consulting contracts, which came through connections from his time abroad. A few short consulting jobs followed. One took him to Belgium for three months.
Eventually, one of these contacts led to a longer-running engagement with a Silicon Valley company.
That engagement is where the company (Kaz Software) started, though nobody planned it that way. The Silicon Valley client liked the work enough to offer Wahid an H1-B visa and a full-time job in the US. He turned it down. He'd just engineered his way back to Bangladesh; moving to America defeated the point. "I said I had come back from England to stay in the country for these reasons. The story has not changed, and I want to stay in the country," he told them. "I suggested to them that what if I hire a team around me, and do their work from Bangladesh."
They said yes. Kaz Software started in June 2004 on the strength of that single six-month contract. Though "Kaz Software" was not yet its name, and would not be for some time.
Wahid wasn't alone when he finally left BdCom. He told Sumon bhai he had an opportunity he wanted to take, and wanted to leave BdCom. When he finally left, two more colleagues from BdCom came with him. "It wasn't fair to BdCom," he says, "but it wasn't really my choice. I mean, they said they wanted to work with me and build this thing together." All three were taking the same bet for the same underlying reason: not dissatisfaction with their jobs, but a hunger for more demanding, more satisfying technical work than a domestic ISP could offer them. "We were paid well at BdCom, but we craved more satisfaction from our work," Wahid says. "We wanted an opportunity to work on software projects to their fullest potential."
Wahid is explicit that this was not a risky bet, by his own accounting. "It was quite a safe entry into this because we had a project," he says. "It was not like I was risking a lot of things; we had a project that was for six months. There was a chance that after six months I wouldn't have work, but I knew that in the software world, you could always find work."
He had also quietly built himself a second fallback behind the first. From savings made on earlier consulting work, he'd been turning over an idea for a short-term rental platform, apartment owners renting out flats for short stays, conceived years before Airbnb existed, and had set aside two or three months to build it if the six-month contract didn’t extend beyond that period. He says, "What I mean is that I had a plan B as well."
The downside case was survivable on at least two different paths, not existential on any of them. The bigger uncertainty wasn't whether Kaz would have revenue on day one. It started with a paying client. It was what would happen after month six. This made the risk-taking more tolerable.
What came after month six was a sequence of small, compounding accidents that look retrospectively like strategy and weren't.
The original Silicon Valley contract went well, so the client kept sending work. Then the client's data partner, a separate company, became a Kaz client too, in 2005, and is still Kaz's oldest client twenty years later. Then that partner's CTO left to start his own company, and he started sending Kaz work as well. The client roster grew by referral chains running through individual people who changed jobs, not through outbound sales or a stated market thesis.
Bangladesh's software outsourcing industry barely existed as a category in 2004. By Wahid's count, no more than five to seven companies in the country were doing genuinely good work in the space, most of them struggling, most no larger than 20 to 25 people. Wahid wasn't executing a plan to professionalize the industry; he was taking the next contract that showed up, from people he'd already worked with.
There was, for those first six months, no company in any formal sense — no registered entity, no office, no name beyond a placeholder. Wahid and his two former BdCom colleagues worked out of his own bedroom. "We were still working out of my bedroom," he says. "When we started the business, I moved my bedroom to our guest room and converted it into our temporary office. My wife was a little annoyed, but such is life." A few tables and a few computers were the entire physical footprint of what would become a 20-year company. He was, at the same time, the only person doing the actual administrative work the operation needed — accounts, office amenities, anything that broke. "If the A/C didn't work, I had to take care of it," he says. "I handled office amenities and environment, and ensured people got paid on time. It wasn't efficient for me to do so many things." He kept doing them anyway, because there was no one else yet to hand them to.
What filled the rest of those six months was work, and a specific kind of pressure behind it. The Silicon Valley client had never tried running a distributed team out of Bangladesh before, and was openly unsure whether the arrangement could work at all — coordinating delivery across that many time zones, with a country that, in software terms, barely existed on the client's map, was an untested concept for them.
Wahid and his small team understood that the six months weren't simply about shipping a project; they were about settling an open question of credibility on both ends at once. "We were trying to convince our foreign clients as well as ourselves that we could deliver world-class software, from design to delivery to deployment to customer support," he says. And he was doing customer support himself, alongside writing code, at the same time.
There was no separation yet between the people building the product and the people answering for it when something broke. The deadline mattered for two audiences simultaneously. The client deciding whether to keep paying for Bangladesh-based work, and the small founding team deciding, mostly to themselves, whether they actually belonged in the industry they'd just stepped into.
The bet paid off on exactly the terms it was made. The six-month deadline was met. The client's own business had improved in the interim, giving it more budget to spend, and Wahid's team had given it a specific reason to spend that budget on Dhaka rather than anywhere else. "Luckily, we met that six-month deadline and things just got better," Wahid says. "The startup we were working for made its first sales, secured some additional funding, and that gave us a bit more runway. Our work also convinced them that this team from Bangladesh could get the job done."
The two events, the client's funding and Kaz's delivery record, came together, but only one of them was under Wahid's control, and it's the one he points to: the proof of work itself is what converted a skeptical client willing to test an unfamiliar arrangement into a client willing to expand it. "When they saw the initial results, they said they would extend the work and expand the team as well."
The referral chain that built Kaz's first ten or twelve clients — the data partner, the departing CTO, and everyone after them — starts at this single point of proof, not at the original contract itself. The contract only bought six months of attention. The delivery is what turned that attention into a reputation worth passing along.
Once the six-month deadline was met, and the Silicon Valley client extended and expanded the engagement, the group decided it needed a more durable structure. "We thought we needed to give it structure," Wahid says. "That's when we formed the company Kaz Software and thus found out that becoming a company means you just added a long list of additional tasks to your to-do list."
Registration happened at the beginning of 2005, and even that simple administrative step came with its own small comedy of errors. Wahid went to the Registrar of Joint Stock Companies (RJSC) office in Motijheel, found a broker, and tried to register the name "Kaz IT Resources." The name search came back clear, he applied, and the broker told him to expect a call in two to three weeks. The call never came. Wahid was buried in deliverables and forgot to follow up, and by the time he returned to finish the registration, a second broker ran the same name search and found it unavailable. It had been taken. By him. "The government system doesn't update properly, and I didn't have any papers proving that it was me who owned the name," he says. "I lost the name to myself. It was so strange and quite frustrating."
Rather than fight the bureaucracy, he chose a new name on the spot: Kaz Software, partly because he wanted something short and Bengali, and partly, he admits, because Kazaa, the peer-to-peer file-sharing network, was culturally inescapable at the time. "That perhaps also influenced me a bit: Kaz to Kazaa."
However, the old name, Kaz IT Resources, lingered as residue for years afterward; Kaz's job postings on BDJobs ran for a long time under the email address kitresources@gmail.com, a fossil of the registration the government had quietly erased.
The company's other piece of infrastructure, its domain, came with its own friction. Bangladesh's .bd top-level domain existed in 2005 but required an in-person visit to the T&T office and a paper form. After "some effort," kaz.com.bd — a clean three-letter domain at the country level — became available, and Kaz has held it ever since.
None of this happened in anything resembling comfort. As the company grew past its first handful of hires, it moved out of Choudhury's house into a rented floor behind a market called Lorna Complex in Eskaton, reachable only by walking through a narrow passage past a row of shops, then across an unexpected open field, to a house where the landlord lived upstairs and rented the lower floor to Kaz.
There was no nameplate, no signage of any kind, ever — not even on the entrance door. Job candidates routinely failed to find the place, and Wahid, who commuted by bicycle and was frequently late, once took a panicked call from a candidate standing outside an unmarked door; by the time Wahid, out of breath from cycling, explained he was five minutes away, the candidate had concluded he'd wandered into a scam and hung up.
Wahid turned the difficulty into a quiet hiring filter rather than fixing it outright: he wrote out a precise, multi-step set of directions — down a certain lane, a certain number of steps, up a particular staircase, knock three times — and treated successfully following it as the first, unstated round of the interview. "Those who could reach this point were the ones who could follow basic instructions and had an algorithmic mindset that we wanted in our people," he says.
The job postings themselves carried out the same filtering logic on a different axis. In most instances, Wahid wrote them himself, deliberately short, about a paragraph, and deliberately indifferent to the credentials most job ads lead with. "We don't care about your degrees or certificates; we only care about your coding skills," ran the gist of it, alongside something closer to a dare: that if a candidate believed they could build Google given the right team, they were right for Kaz. No paper contracts were issued; hires were made by email, on the strength of a stated word. "If this level of trust doesn't exist in a group of people, you can't work together," he says of the no-contract practice, which Kaz still keeps even at a much larger size today.
Every piece of the early operation, the inaccessible office, the deliberately direct but ambitious in its own way job ad, the handshake-by-email hire — selected, more by accident than design, for the same kind of person: someone willing to take a risk on an unmarked door in an odd part of Eskaton, working for an oddly named company nobody had heard of, on the word of a man who was usually a few minutes late.
Twenty years later, Kaz is one of the more durable software outsourcing firms in Dhaka, and is known in Bangladesh's tech industry less for its size than for an unusual property. A disproportionate number of its alumni have gone on to found their own companies or take senior leadership roles elsewhere.
Wahid attributes this to a deliberately built culture of trust and minimal hierarchy. Informal hiring focused on problem-solving and collaboration rather than credentialed knowledge, no rulebook, no monitoring of how employees use their time, an "underground" peer leadership structure ("the Mafia") that vets management decisions before they're rolled out, and an explicit insistence that the workplace be a place where people are happy, because you only get one life and most of its waking hours go to work.
None of that culture-building happened before the company existed. It was layered on after a six-month contract turned into something that needed a second hire, then a tenth, then more.
A note on what to watch for in this case
Four things in this case are worth holding onto as you read other starting-business cases.
The first is the relationship between the founder's stated risk tolerance and the actual risk of the decision. Wahid frames his own personality as low-ambition, almost reluctant. Before recounting how Kaz started, he offers a warning: "People who know me, if you ask around, the biggest complaint about me is that 'bhai, you have no ambition.'" He says he is driven, "but not for the obvious things" — money, he says, "never drives me." Yet the company started on the back of an already-secured contract from an already-known client, which he calls, in his own words, "quite a safe entry." These descriptions may feel like Wahid actually didn’t take a risk. But that explanation is more about perspective and risk tolerance.
In a sense, Wahid's journey can be seen as a pattern of taking the asymmetric bet that's already on the table. Leaving the PhD because of a closure-based work preference, plus a part-time job offer in hand; leaving London because of family reasons, plus arranging a job in Dhaka before returning. Wahid had always taken risks throughout his career. In fact, he took quite consequential risks throughout his journey. The good thing was that he always tried to figure out what would happen if things didn’t work out as planned. He always tried to engineer a fallback. In fact, when he started the Kaz project, he created a second project to fall back on if the client work didn’t continue after the initial six-month contract. Similarly, Wahid has a stoic personality. He has a rather detached approach to the world. This is also visible in how he perceives risks. To many people, which is a paralyzing uncertainty, is something figureoutable to Wahid. He knew starting a business was risky. But he also knew if it didn’t work, he could go back to working for another company or start a different business and he was comfortable with thoe alternatives.
The second is the role of relationship debt in the absence of a market. Bangladesh had no real software outsourcing industry to plug into in 2004, no established sales channel, and no comparable firm to benchmark against. What Kaz had instead was a chain of individual people — a physics department senior who ran an ISP, a foreign colleague who made an introduction, a client's data partner, that partner's departing CTO — each act of trust feeding the next. None of this shows up in a business plan. It only becomes visible afterward, when you ask a founder to narrate the chain.
The third is the gap between when a company starts and when it becomes a company, and how little the gap matters in hindsight. By any formal measure — registration, a name, a bank account, a lease — Kaz Software did not exist for the first six to eight months of its own founding story. What existed was three people, a contract, and a converted bedroom. The decision to formalize came only once the original bet had already paid off once (the contract was extended) and the administrative load of staying informal had become heavier than the cost of dealing with the RJSC.
This suggests a sequencing worth checking against other cases: structure, in this telling, isn't something a business needs to start; it's something a business needs once continuing to operate without it becomes more expensive than the paperwork. Even then, the registration itself was handled with the same loose, reactive style as everything before it — forgotten mid-process, then re-attempted under a different name almost by accident.
The fourth is the distinction between what bought Kaz its first six months and what actually extended them. The contract was the entry ticket; it explains why Wahid could afford to take the risk at all, and it's the detail that makes the "safe entry" framing in the first note credible. But the contract alone explains nothing about why a single client engagement turned into a twenty-year company with a multi-generational alumni network.
That second thing is explained by a different mechanism entirely: a small team treating an already-secured six-month contract as if it were still fully contested, because in the client's mind, it was. The client wasn't simply waiting out a contract term; it was running a live test of whether a country it had no prior model for could deliver work it could trust.
Wahid's account makes clear that he understood the test was running, on both ends, simultaneously. Convincing the client and convincing the team itself were, in his telling, the same exercise.
Kaz's later culture is built on the principle that trust, once extended, doesn't need to be re-earned through visible effort. But trust had to be earned once, at the very beginning, through exactly the kind of visible, demonstrated effort the company later stopped requiring of anyone.
The source for this case is our interview with Kaz founder Wahid Choudhury. The case is part of the How Businesses Get Started sequence.
