Hameed Afssari:
Localized Software for Everyone, Everywhere



Shiraz, Iran


Wichita State University


London (UK), Hyderabad (India), Deception Pass (US), Mecca and Medina (Saudi Arabia), and Isfahan (Iran)


Reading, writing, hiking, and soccer

These days, software is essential to how we get things done and interact with one another, with certain apps and programs used ubiquitously throughout the world. Having led localization and internationalization teams at two global software companies — first Microsoft and now Uber — Hameed Afssari believes that software products should be easy to use for everyone, no matter what language they speak or where they live.

For an app like Uber with a truly global reach, localization involves plenty of creativity — something that Afssari embraces when navigating the complexities of myriad languages, cultures, and government regulations in thousands of cities worldwide. Through process and technology innovations like “touchless translation,” Afssari’s team has managed to significantly increase the amount of content it localizes while keeping costs low.

Afssari sees internationalization both as a “growth-enabler” that brings in more customers and — equally important — as a means of helping people, whether it’s adding Pashto so that Afghan refugees can use the platform or allowing riders in majority-Hispanic US cities to request a Spanish-speaking driver. Based on a quarter century of experience in the industry, Afssari shared with us his advice for staying ahead of the curve and why he’s optimistic about the future.


Can you tell us about your background, including your education and exposure to languages?

That’s an interesting story. I’m from Iran, but moved to Kuwait when I was seven. I went to an Iranian school there, so I was exposed to both Persian and Arabic, and some other languages like English and a bit of Urdu.

When I finished high school in 1986, I wanted to study in Europe, and Yugoslavia at the time was a choice for many because of the low cost of living. The Serbian language was a bit difficult, and the script was Cyrillic. I ran out of money after six months, so I went back to Kuwait and worked for three years, determined to come to the US this time. When I eventually did in 1989, I went to — of all places — Wichita, Kansas. I did my bachelor’s in computer science there and then my MBA in management information systems.

When I talk to people, they can’t figure out my Arabic accent. They ask which Arab country I’m from, and I say none — I’m Iranian. So it’s a good conversation starter. And I think it helped me land my first job after finishing my master’s: a contractor position at Microsoft testing the Office suite in the Arabic language.

The first day on the job, they said, “We’ve hired too many Arabic speakers, but we want you to test Thai and Hindi.” I told them I don’t know those languages, but they said, “It’s not linguistic testing; you’re doing functionality and globalization testing.” It didn’t take long for me to become a full-time employee, and shortly after that I started leading internationalization testing for these languages and managing the international automation and configuration lab.

Later in your career at Microsoft, you led the “Office World Readiness team.” What was the idea behind this team and what were your primary responsibilities?

As Microsoft Office added more apps — such as Outlook, Infopath, Sharepoint, and Skype — it was not scalable to keep hiring testers to test these apps in different languages. We needed to teach developers that the software has to be global by design, so that bugs are not introduced in the first place. The idea was to create processes and best practices to make the developer aware of globalization from the start. Our approach shifted from reactive to proactive, and we built tools such as a i18n code reviewer and dynamic pseudo-localization to move globalization quality upstream.

My job was half testing and half education. We created incentives and courses for anyone working on the development side so they could create software that is ready to be globally shipped. The idea of simultaneous shipment, or “sim-ship,” came about 15 or so years ago. That means the Microsoft Office English version is going to be available at the same time as the non-English versions for the rest of the markets.

I think that idea has caught on with other companies, who now say “global first.” At Uber, we have a global-first strategy, meaning you can’t wait to ship a feature in other languages or markets. It has to be ready to go everywhere at the same time.

During your time at Microsoft, many non-commercially viable languages were added to the Office suite. Why was it important to include those languages?

There was a group of folks at Microsoft, including myself, thinking about this that said, “What if we can localize — in a very affordable way — most of the UI? And use that to save languages and also enter into new markets?”

One goal was to promote and evangelize computer literacy and save endangered languages, such as Cherokee in the US. It was a bit difficult, and there was pushback, but we were able to persuade management that this is a good idea.

Another goal was to tap into new and emerging markets like India, where we could support additional languages. For Farsi, or — as some folks like to call it — Persian, Microsoft had a version way before, but because of sanctions on Iran, they couldn’t build it anymore. So when we said we wanted to do Farsi again, they said, “We are not doing any business with Iran.” In response, I argued that Persian is spoken in other countries like Tajikistan and Afghanistan, and there are big contingents of Persian-speaking expatriates in other countries.

They were convinced, but then the challenge was where to find folks who could localize the Office UI into Farsi. The University of Washington had a Farsi language program. What if we used students in this program to help localize? We hired four ladies who were native speakers and had also taken Persian literature courses there. They localized Office, and the quality was really good.

After a brief tenure at Workday, you landed at Uber and soon took on the role of head of globalization. What attracted you to Uber as a company or inspired you about the work to be done?

I have been fortunate to work at three great companies in my 25+ year career, and I joined these companies at different stages of their growth. When the position to lead the Uber localization operation team opened up, I was looking for an opportunity for more immediate impact and creativity. I loved the challenge detailed in the job description. The team had had a couple of leaders leave in a short time span, and I think they were yearning for stability and growth.

When I joined, I made a couple of important changes in close collaboration with other leads and senior folks in the team. One thing was how we used vendors. At the time, it was one vendor doing translation and another doing the quality checking. The model didn’t make sense to me; if you’re doing work, you should be responsible for its quality. This was the same idea that we had at Microsoft: developers should be responsible for quality as much as test. So, we may have a multi-vender strategy, but that doesn’t mean that the work goes back and forth between two vendors.

The second change was bringing in machine translation (MT), which received full backing from leadership. So the changes I brought in with the help of the team were transformational in essence.

With Uber operating in more than 70 countries and 10,000 cities, what’s the most challenging part of adapting the software to such a wide range of locales?

One of the biggest challenges is that the regulatory environment is very local and very active, so our translation model must be highly customizable and agile. Due to legal requirements that may vary from state to state or even city to city, we may end up using different source terms, and that becomes very challenging as it defies the basic idea of one source term translated into many target languages.

For example, we may end up using a different term for what is otherwise known as a “driver.” In the US, the term is “earner” on the Uber platform. In other markets, we may use “operator” or “conductor” as the source term to produce the legally approved translation. We have to have an automated process for saying, “In this language, we can say driver, but in this other market, we’re going to call them operators or conductors.” How do we do this in a seamless way, making sure that when we switch the term the sentence structure is correct?

Additionally, cities or states may pass mandates that say, “Most of the earners on your platform in the city of Seattle, for example, are not even native English speakers. We want you to support Somali and Eritrean.” Even though Uber is not in Somalia and Eritrea, we can offer these languages. So we are serving the people in their own language, so long as they choose to use it.

Because it’s not commercially viable to translate everything through humans, we may use MT-only for some content. We also decide which model will work with a given language, for example. This all contributes to the creativity of the teams and how they want to solve a challenge.

One important fact to call out about the Uber localization team is that we have our own data team, which we rely on to look at who’s using the app, how and where they’re using it, what features are getting the most usage, and what content is leveraged the most. They tell us that there’s an opportunity to support this language here, or that not that many users are using this language there. In short, the data team has helped us make data-driven decisions when it comes to supporting a language or market.

What is the driving force behind the Uber localization team, and what are some of the projects or accomplishments you’re most proud of?

Our vision is to make Uber feel local to everyone, everywhere and to provide a world-class user experience across the board at scale. This means looking at localization not just as a transactional process to translate specific content. It means thinking globally from ideation all the way to engaging with users when a feature is released.

One notable accomplishment is that we have reduced our cost per word (CPW) by more than 50% while our volume has almost tripled and our quality has remained high, as confirmed by very low error rates and user satisfaction surveys conducted by Nimdzi. We were able to do this with the same budget for localization by creating more strategic partnerships with our vendors and by deploying touchless translation in which no human is involved (this year, 70% of our content is going to touchless translation).

This opened up opportunities for us to go beyond localization of our content into localization of user-generated content, like if the driver and rider are talking to each other and the languages are different. Let’s say you are in Amsterdam and you say to the driver, “I’m running five minutes late. Can you wait five more minutes for me to get out of the building?” You type in English. His device language is set to Dutch, so he will see it in Dutch. We want to make communication seamless; there should be no friction there.

One of the questions Nimdzi asked in its survey was, “If Uber were a person, where would you think they came from?” If someone using Uber in Saudi Arabia says it’s from the US, that means we have not done a good job. But if they say it’s from Saudi Arabia or Egypt, or one of the Arabic-speaking countries, then you have been successful.

We’ve expanded these scenarios one by one, and have extended it to support calls. Do you need multilingual people for this? What if we can hire someone whose only language is English? This is the type of work that has raised awareness about our team. These are not just translators who don’t know anything about tech — these are folks who are enabling the growth of Uber. In fact, this year, other teams at Uber nominated localization folks for the Reimagine Award, an internal recognition for employees who make a significant impact on the business.

I am proud of the fantastic localization and internationalization teams we have at Uber and how they break many barriers and build bridges that reinforce Uber cultural values day in and day out. The team is very diverse geographically and culturally, which brings diversity of thought to whatever we want to build and how we approach our stakeholders.

We emphasize that language is an accessibility feature, similar to accessibility features that are mandated in the US and some other countries. If you globalize your code, instead of 1 million people in the US using it, 100 million around the whole world will have access to it. Hopefully other companies can also benefit from that mentality and use it to make a case for their team.


When it comes to purchasing products and services from language service providers (LSPs), what factors do companies like Uber look for, besides the obvious ones like cost?

There are two major factors that will help with the cost, quality, and speed. One is tech-readiness, meaning a provider can scale, adapt, and innovate to keep up with increasing demand and complexities of our content. You’ve got to be ready from day one to enable integration with our environment. The idea of “we will build it when they come” doesn’t work for us. We have to see that you’re capable of doing the work right now. It is very important to be able to showcase that to us.

The second one is your “customer obsession” — making things happen for the customer. How quickly are you able to fulfill the needs of your customers? Do you say, “Yes, we’re going to do it,” but it takes you six months? We have this as a value at Uber, and we expect the same thing from vendors. The customer obsession goes both ways: how you treat us and how you treat the people who are translating for you. I hear horror stories that people get paid a few months later. They have translated for an LSP but haven’t gotten paid. These are the things that are very important to us.

Is there anything else you would like to add?

I want to emphasize that the future of localization is very bright. There is a fear in the market that large language models (LLMs) are going to replace the whole industry. From what I’ve seen, actually, LLMs will enable growth in the industry. LLMs need skilled people to define quality frameworks, verify quality, and enhance learning capabilities. There may be a shift in terms of which team within the broader localization team will be busier; whichever team is responsible for linguistic and localization quality will see a lot more work. But the bottom line is that the localization industry’s future is now.

To take advantage of this, you have to be tech-ready. You have to use the technology to your advantage to bring your costs down without impacting the bottom line negatively for your translators and linguists. We are taking those measures to make sure we are ready at Uber, and also to help any other company that needs to go global, scale its quality assurance, or improve its AI capabilities. Growing internationally should be a mutual goal for all of us because we are one big family.

Cathy Martin is managing editor of MultiLingual magazine.


Related Articles

Apple, Byte Level Research, Microsoft, Nestle, wikipedia
pen on paper

The 25 Best Globalized Websites

By MultiLingual Staff

After recently releasing his Web Globalization Report Card for the 17th year, author John Yunker said he’s been getting positive feedback on what it signals…

→ Continue Reading