When Raz Besaleli isn’t on the Montclair State University campus participating in the Natural Language Processing (NLP) Lab, tutoring fellow linguistics students or helping professors with their courses by grading assignments and attending office hours, she’s working on her master’s in Computational Linguistics – or helping run a multimillion-dollar AI company she helped co-found.
It’s all in a day’s work for Besaleli, a founding NLP research scientist at Seek AI, which recently landed $7.5 million in pre-seed and seed funding, according to a company press release. Besaleli estimates that on a weekly basis, she puts in 60 hours at her job and 20 at Montclair. “I don’t get much sleep but that’s OK, I’m doing what I love,” she says.
Seek AI, Besaleli says, provides “a natural language interface for a company’s data. The idea is that you should be able to ask any question that you want, in natural language, and it will pull the data for you from your databases. The motivation behind this is that companies often have to either hire a lot of data analysts to just write this code called structured query language or SQL to pull this data from what are often very messy databases.”
“They often find a lot of the time consumed writing something that we believe could be automated,” Besaleli says. “So, at the core of it, we are building code-generation models that will be able to generate code, given natural language that will interact with your database for you. We’re streamlining the process of finding answers at a company and streamlining the way that it’s done.”
Results, she says, could be complete in minutes versus days. Still, Besaleli says, Seek AI’s objective is not to replace data analysts. “Our goal is to make their jobs easier so that they can do more meaningful work,” she says.
Besaleli adds that “The generative AI space is very, very new. Even six months ago, we didn’t have the capacity to do what we’re doing now. It’s a very, very rapidly growing field.”
Already, the SoHo-based Seek AI has more than a dozen pilots with tech, fintech and consumer brands customers ranging from startups to Fortune 100 companies. Seek AI will use the funding infusion to “accelerate product development and support its mission of making data accessible to anyone in an organization via a natural language interface,” the press release states.
As a co-founder, Besaleli played a role in obtaining that funding. “Raz would do investor pitches with me and speak about the underlying technology behind Seek,” says Seek AI Founder and CEO Sarah Nagy.
Nagy says she had no qualms about Besaleli still being a graduate student when they started working together. “I was really impressed by Raz’s background in linguistics, combined with her technical chops working with transformer models in PyTorch [an open source machine-learning framework that accelerates research prototyping to production deployment]. She also had co-founded a startup before, in the aerospace industry, which was incredibly impressive. On top of all that, she had a great attitude: eager to learn, get her hands dirty, and work hard to build Seek. Once I had known her for a few weeks, I knew I had to get her in the boat.”
At the time she and Besaleli met, Nagy says, “I had built a very small prototype of an AI model and was facing the challenge of scaling the model and getting it to work with real-world data. Raz came up with several innovations that solved these challenges and helped propel Seek to be the leader in our space.”
While ChatGPT has entered the artificial intelligence lexicon thanks to tons of media attention, Besaleli says there’s a big difference between that tool and the capabilities of Seek AI. “We use a model like ChatGPT in our systems, but the scope of our systems are much, much larger. ChatGPT is just a model, and it can generate SQL but from a very, very limited scope,” she says, adding, “What we do is a little different because we’re also a repository, a centralized location for all the questions that that business asks. We’re very much focused on information retrieval, not just generating code.”
Besaleli is already an expert on the subject. She recently was the only student alongside three Montclair professors on the faculty panel “ChatGPT and the Future of AI” at the Feliciano School of Business. She sat alongside her research advisor and Computer Science Professor Jing Peng, Linguistics department chair and Computer Science Professor Anna Feldman, and Writing Studies Associate Professor Ronald Brooks Jr.
When Besaleli was weighing her options about which school to attend, the Annapolis, Maryland, native looked at universities in the U.S., as well as the University of Reykjavik in Iceland. Ultimately, she chose to study at Montclair largely because of Feldman. “I was interested in her research, and we just got along really well.” Besaleli adds that affordability also was a factor in her decision.
Feldman says she was unaware that she played a role in Besaleli’s choosing to attend graduate school in New Jersey. “But I’m happy Raz joined our program,” she says. “We really enjoy having her at Montclair.”
Montclair is one of the few universities in the country that offer a Master of Science in Computational Linguistics – and the only university in New Jersey. “It’s also special because it’s truly interdisciplinary, housed in both Linguistics and Computer Science, and gives our students a strong background in both computer science and linguistics,” says Feldman.
Besaleli, Feldman adds, is an excellent student. “She never takes any facts for granted; she always wants to challenge ideas and verify claims. She worked with me on a project trying to discover historical semantic shifts in Modern Hebrew and discovered very interesting patterns,” she says.
With the guidance of Peng, Feldman says Besaleli is “exploring how large language models can translate SQL queries into plain English. [SQL is a domain-specific language used for managing data held in relational databases, and SQL queries have a specific structure.] Raz would like to find out whether generating an SQL query from a text description requires a lot more context than the other way around.”
As for the future, Besaleli expects to graduate in May “if everything works out and if I have time to write my final projects,” she says, laughing. She certainly doesn’t have to worry about the job market, and she’ll be able to devote her attention to Seek AI.
“I see this revolutionizing the way that people interact with their databases,” she says, “and I hope that we can just make life as easy as possible for those in the data space, so that they can do the things that are important and that they want to do.”
Feldman says the Seek AI platform “is a great product that uses sophisticated natural language processing techniques and machine learning to allow users to use natural language commands to query databases…. As for Raz’s future, life is an adventure, and I hope Raz will continue building AI applications for social good.”
Meanwhile, Nagy is looking forward to navigating Seek AI’s future alongside her co-founder. “Raz is, by many accounts, a genius,” says Nagy, “and I am excited to continue to work with her.”
You may also like:
Why All the Chatter About ChatGPT?