☰ Menu eFinancialCareers

J.P. Morgan’s ex-Microsoft head of machine learning quit for Facebook (updated)

Geoffrey Zweig Facebook

Remember Geoffrey Zweig? The natural language processing (NLP) expert J.P. Morgan brought in from Microsoft to be its global head of machine learning? We hear that he’s gone after little more than a year. We understand that he’s gone to Facebook.

J.P. Morgan declined to comment on the claims that Zweig quit after receiving his bonus and Zweig didn’t respond to attempts to contact him on LinkedIn. When we called his number at J.P. Morgan in New York, the phone went unanswered.

Zweig’s exit is a blow, not to just to J.P. Morgan, but to every bank looking to hire-in hard-to-find machine learning talent and then keep it. Zweig is a big name: he spent over 10 years at Microsoft and his team there was credited with making a major breakthrough in speech recognition systems. As banks seek to develop artificial intelligence programs that can read documents and potentially interact with clients, they need more people of Zweig’s calibre.

It’s unfortunate then, that as Amazon discovered when it tried to build a machine learning team to create Alexa, people like Zweig don’t always want to work in the “commercial” sector, preferring to remain in academia or more research-oriented teams at Microsoft or Google. Since joining J.P. Morgan, Google Scholar indicates that Zweig’s research output declined to a trickle, with his name included in just seven published articles or patents last year, compared to hundreds the year before while he was still at Microsoft. As part of Facebook’s research team, he’ll be back at forefront of new developments.

J.P. Morgan is typically credited with being more advanced than other banks in terms of machine learning. Late last year it rolled out LOXM, a self-teaching trading algorithm that can execute the kind of large, complicated equities trades usually reserved for human beings. Daniel Ciment, the bank’s global head of equities electronic trading, is a big fan. However, there appears to have been some confusion regarding the bank’s AI strategy: Samik Chandarana, a former credit trader and J.P. Morgan veteran, was made head of data science and analytics in October 2017 and became the spokesperson for J.P.M’s machine learning strategy, effectively sidelining Zweig.

Insiders claim Zweig and his team have been bored and that J.P. Morgan didn’t have the systems in place to make the most of their abilities. With Zweig apparently gone, the bank is said to be looking to build out its AI capabilities under Chandarana instead.

Have a confidential story, tip, or comment you’d like to share? Contact: sbutcher@efinancialcareers.com
Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libelous (in which case it won’t.)

Comments (2)

  1. Bank management used to think hiring more tech and quant could bring more innovation yet only resulted in inefficiency. One example is “central risk book”. A few years ago, it was the buzz word on the street. However, almost every bank’s head of central risk book left last year. Due to regulation and market structure change, it is going to be hard for this desk to have good p&l. What banks should really do is to keep this team small and efficient, rather than keep adding people and creating redundancy.

  2. Agree. Central risk book traders at my firm are discussing how to create prop trading strategies by reverse engineering client order data. They want to start their own hedge fund with these strategies.

The comment is under moderation. It will appear shortly.


Screen Name


Consult our community guidelines here