Safaba

Alon Lavie

Pittsburgh, Pennsylvania
Cofounder & CTO, Safaba (Acquired by Amazon)

Safaba is an automated translation solution for global enterprises’ digital content, like websites, customer communications and more.

I spent most of my career in academia as a professor at Carnegie Mellon University (CMU) in the AI language technology space. In 2009, a cofounder and I started Safaba, which was essentially a spin-out of my research lab activities. Automated translation technology was evolving quickly at that time, mostly in research settings, and we identified an opportunity and need for commercialization. The technology was particularly advantageous for large, global enterprises—including Amazon, who initially approached us as a customer and really liked the product and expertise we offered.

A lot of our technology development was funded by the Small Business Innovation Research grant program and some investment capital from a few different entities. By the end of 2014, we were at the point of raising a series A round when Amazon approached us with an offer, and we ultimately decided to accept. We were a team of 12 at that point—mostly from the CMU research environment—and eight of us went full time into Amazon. At the time, Amazon was the only large tech company without a presence in Pittsburgh and we made clear early in the process that none of us wanted to move to Seattle. I convinced them that there is a lot of value to the talent and connections at the University, so as part of the acquisition they opened a corporate office here in Pittsburgh. I worked there as a senior manager for three and a half years post-acquisition, and the office has grown to 300-400 people, so I definitely pat myself on the back for being the person recognized as bringing Amazon R&D to Pittsburgh.

Focusing on the positives, I think being acquired was a really good outcome for Safaba. Ultimately, it was the right decision for the company, and it wasn’t just financially lucrative for my cofounder and I, but it was a transformational professional opportunity and financial outcome for our entire team. The integration of the technology within Amazon went really well. And managing an R&D team in a large company also added another highly valuable chapter of experience to my career that I really appreciate.

Being acquired is not without its challenges, though, especially with a large company like Amazon. Integration from a culture standpoint is really tough and generates a lot of situations for people to become unhappy. We were a small startup with roots in the University research space, and so the transition to a large company like Amazon was difficult in terms of operational structure, rules and how organizations are managed at that scale. When it came to the acquisition negotiation process, we were also on a different playing field in terms of resources and experience, and so that was probably the biggest challenge for us as founders. Even where we did things right—we had done an immaculate job of clearly separating and documenting our IP in the technology transfer process with the University— there was friction with Amazon. We also had a long relationship with a top-tier legal team that we weren’t able to leverage because they represented Amazon elsewhere, and Amazon wouldn’t give a waiver to allow our legal team to represent us in the transaction.

Ultimately, I’m glad we saw the acquisition through, but I think there’s a lot policymakers, startup supporters, and others can do to help empower startups in the acquisition process, particularly when the acquirer is a large company like Amazon. Shared tools and resources seem like a good place to start. Template agreements or standard terms might help founders understand what is standard in a contract, but wouldn’t be very valuable if the acquiring companies are able to toss them aside in negotiations. And for startups, high-quality legal and business representation that you trust to negotiate on your behalf is critically important, as is ensuring your proprietary IP is clearly identified and welldocumented to avoid the potential for issues in the acquisition process.

Today, I am an adjunct professor at CMU and a senior manager at a bi-national growth-stage scaleup called Unbabel. Unbabel is fundamentally in a similar AI translation technology space as Safaba, providing an AI-based platform for translation of large volumes of multilingual content for large enterprises. I knew the founders long before they actually started the company—the CEO actually got his Ph.D. at CMU. I opened an office for them here in Pittsburgh and largely oversee the AI technology side for the company. As a growth-stage scaleup, Unbabel is another interesting chapter in my career in the translation technology and NLP R&D space that rounds out my experiences outside of academia in terms of both founding and running a small startup and working at a large tech company.