When Amazon acquired Alon Lavie’s automated text translation service, Safaba, in 2015, he had one condition: the startup’s team would remain in Pittsburgh.
“... If you want to acquire Safaba, you have to come here,” Mr. Lavie recalls telling the mammoth e-retailer headed by Jeff Bezos. “We’re not moving to Seattle, we’re not moving anywhere else ... you have to open an office here.”
Fast forward to today and Amazon’s South Side office, which is responsible for much of the underlying voice technology and machine translation used in its Alexa voice assistant, has over 50 employees.
Mr. Lavie is no longer one of them.
Formerly the senior manager for machine translation at Amazon, he’s moving into a new role as vice president of language technologies at a company called Unbabel.
That company, which has technical headquarters in Lisbon, Portugal, and business offices in San Francisco, focuses on language translation services for businesses.
Read More
Once again, Mr. Lavie has a similar provision: he will remain in Pittsburgh and he will build a technical team — here — in the next six to 12 months.
“I didn’t have to convince them at all ... why expanding in Pittsburgh is a wonderful idea,” he said.
That’s due to the city’s widening reputation as a home for language technologies.
The city’s most highly valued home-grown startup likely is Duolingo, the East Liberty-based language learning company that uses gamification in its online lessons. As of July 2017, it was valued at $700 million.
WeSpeke, a language learning website and app meant to connect users by their common interests in order to spur conversation, has partnered with CNN to use news stories as lesson content. That startup has raised at least $6.2 million, according to Crunchbase, an online portal where firms self-report rounds of funding.
Not to mention, Carnegie Mellon University’s Language Technologies Institute has been gaining clout since it was established in 1996 as a division of the renowned School of Computer Science.
Alan Black, a professor at the institute, told the Pittsburgh Post-Gazette in April 2018 that it’s the largest program of its kind in the country, with at least 350 graduate students. He said that upon graduation about half will move on to companies like Amazon, Google and Apple.
Unbabel, meanwhile, could be the next frontier for Pittsburgh’s expanding language technology field.
The company uses artificial intelligence and machine translation to help companies with customer support in international countries.
For example, enterprise-level companies must often cross language barriers across the firm’s offices to make content available to all of its employees and customers.
Up until five or 10 years ago, companies traditionally recruited professional translators to decode different types of content for different audiences, with little to no technology, Mr. Lavie said. Each time, they had to start from scratch.
At a point, it made sense for human translators to instead spend time working alongside machines. That is, editing translations that computers come up with to help perfect them.
If you’ve ever copied and pasted a phrase from one language into Google Translate and found the English version a bit odd, that’s where a human editor could be helpful.
Unbabel uses this model, where an artificial intelligence-powered platform creates a translation of some content, but then human editors step in to make it better. The process is known as “post-editing,” Mr. Lavie said, and it eliminates the need to start from scratch time and time again.
To date, Unbabel has received at least $31.2 million in funding, according to Crunchbase.
Mr. Lavie will help build out an even more robust translation platform for these human workers to increase their productivity. That’ll ultimately save money for Unbabel’s clients, which include Facebook, Pinterest and Change.org, a website to create online petitions.
He’ll be doing that remotely from his home in Squirrel Hill.
Before Mr. Lavie can start building a Pittsburgh team, though, he needs to first outline his goals with the teams in Lisbon and San Francisco. Then, he’ll be able to figure out the precise areas where a Pittsburgh team could benefit Unbabel.
He doesn’t know yet what the scope of that team will be, but his passion for keeping language technology in Pittsburgh is no accident.
When he earned his Ph.D. in computer science at CMU in 1995 and was subsequently offered a job to stay around as faculty, he jumped at the chance. At the time, Mr. Lavie said, “The area of language technology was greatly expanding” at the university.
But that alone wasn’t enough to spur commercial activity.
“There were a lot of opportunities because we were growing really top notch graduate students in the space,” Mr. Lavie said. “[But] most of them ended up finishing their studies and leaving for other companies in other big places.”
Through Safaba, Amazon and now Unbabel, he’s starting to move the needle.
“I wanted to help create an opportunity for people to stay in Pittsburgh if they really wanted to stay.”
Courtney Linder: clinder@post-gazette.com or 412-263-1707. Twitter: @LinderPG.




Comments
Welcome to Post-Gazette subscriber-only commenting community. Please review our community guidelines.