At first glance, Keith Lubell looks like he works at a Wall Street investment bank. He dresses in smart suits complete with immaculately tied bowties. He takes his time and speaks carefully and articulately. Lubell fits in well at Berkery Noyes, the downtown mergers and acquisitions (M&A) investment bank where he’s served as its CTO for 15 years.
But looks can be deceiving. Lubell is a West Coast kid from Los Angeles who got his degree from Middlebury College, a small, prestigious liberal arts institution in Vermont, nestled between the bucolic Green Mountains and the Adirondacks. While this might suggest that he is professorial or even a bit of a hippie, his education and upbringing don’t tell the whole story.
After college, Lubell joined the Department of Defense (DoD), creating warfare simulators for B-52 bombers and attack helicopters. If that seems more in line with the bio of a Wall Street executive, before he joined the DoD, he moved to New York to play in a rock-and-roll band. After the DoD he took a few years off and moved to Spain for six years with his wife to play music and paint and create stained glass.
No, Lubell can’t quite be painted into a specific segment, but the platforms he creates do just that—segment and organize companies to help Berkery Noyes’ directors to find investment opportunities.
“Because technology was moving so fast, you could kind of jump in because everyone was still learning new things all the time. There were still mentions of technologies that I had never paid attention to that would’ve been advantageous, but it wasn’t so big because in the technology business you have to learn all the time. You can’t sit on your chops.”
Berkery Noyes was founded in 1983 by Joseph Berkery. The institution provides mergers and acquisitions advisory and financial consulting to middle-market entrepreneurs in the global information software, marketing services and technology sectors.
Even though he’s in his 80s, Joseph Berkery still skis and, impressively, knows his technology companies. Chatting with him at the firm’s One Liberty Plaza offices, the eldest Berkery—his son James is the managing partner—rattles off and asks questions about the major order management systems (OMS) providers in the space and then talks about the merits of certain programming languages for research. Joe knows his tech and he’s fond of Lubell, as well.
“Keith is one of the top 50 physicists in the world; I believe that,” Berkery says with a warm smile.
When you’re covering the technology space, you can’t just talk about innovation—you have to be innovative.
“As a bank, since our clients are innovative entrepreneurs, we have to be innovative, too,” Lubell says. “We have to be a technology-enabled service; we can’t just be a stuffy, old investment bank anymore.”
James “Jim” Berkery, the managing partner, has helped to lead this mindset shift. The younger Berkery served as the CIO and understands the importance of information technology.
“Jim is very interested in technology,” Lubell says. “He saw the need to change the way the bank ran—to change this bank to be a technology-enabled bank. Investment banking has been the old-boy networks. You play golf and go to the downtown club where you’re a member—it’s traditional networking. While there’s still a lot of that since relationships are so important, just doing that is not scalable. So to make a bigger bank, you need to go beyond that and Jim foresaw that—you need to build an infrastructure.”
The Best Defense
Lubell graduated from Middlebury in 1985 with a B.A. in physics and a minor in studio arts. He studied lasers and microprocessors for his major and was a bass player in the college’s various music ensembles to achieve a minor. His rock-and-roll career didn’t last long before he was snatched by the DoD, where he worked as a software engineer, coding in the Department’s Advanced Systems Development outfit. While there, he designed and implemented radar systems on projects for the B1B bomber and the SH-2F anti-submarine helicopter.
The job entailed creating hostile territories and modeling aircrafts going into enemy areas and scoping out how the radars would react to planes and helicopters coming into the environment. These simulators—designed at the tail end of the Cold War—would help pilots to train prospects for changing warfare environments.
While developing war simulators for the DoD might seem like a far cry from working at a buttoned-down investment bank, it was here that Lubell was introduced to semantic technologies and the importance of Agile development—which wasn’t quite a thing yet—that he would take with him later and deploy at Berkery Noyes.
“Some of the innovations that we’ve done here are built around building idea spaces,” he says. “We have a semantic space for modeling the way businesses work and if you look at the landscape of all the businesses out there, you can define a semantic topology—a space where they live—and say, here’s this business in this multi-dimensional space and here’s this business at this point in the space, and there are distances between the two. It’s a mathematical construct. Bankers talk about industry spaces all the time—I’m in the healthcare space or the tech space—but we literally build a mathematical space based on those semantic ideas. I brought that from my time with the DoD, where we modeled an electronic warfare space.”
It’s All Semantics
Understanding the terrain is vital in warfare and in business, even if the results of failure are vastly different. Berkery Noyes’ business is M&A and a key piece of that is being able to map out the sector and create a topography of it. A managing director at Berkery Noyes will have to talk to hundreds of companies before signing one on as a client. The environment has to be right to be a buyer or a seller. Creating time-saving processes and being efficient are the differentiators between striking oil and pulling up to barren land.
Much like creating an aerial warfare simulator, Lubell looked to build database environments that would lay out the various M&A fronts that Berkery Noyes was competing in. This yielded a proprietary system that ties together a customer-relationship management (CRM) platform, an email marketing system, a research database, and a workflow system. It is fully integrated and contains 150,000 companies, 50,000 M&A transactions, and nearly 200,000 people. The result is a microcosm of each sector that Berkery Noyes does business in.
With the multi-dimensional space linked together, a director can look at one company and see all the associated deals and various ways that they compete, allowing Berkery Noyes to show clients comparisons that they might not have thought of as potential buyers or sellers. An individual could try to do this by hand, but Microsoft Excel only has so many columns and rows and the human eye can only connect so many dots. While shadow IT will always exist, these programs are limited.
About 18 months ago, Berkery Noyes began adopting ElasticSearch—an open-source search engine that combines HTTP web interface and JSON documents that use human-readable text to send data objects consisting of key-value pairs—into its database. The bank began feeding all of its data into the engine. The information is stored in document form and is indexed. Embedded in those documents are the platform’s semantic topology elements. As a result, when running an ElasticSearch using filters, the data has been “massaged” into the engine’s topology to produce more targeted results that can better find connections, Lubell says.
He admits that they’re just scratching the surface when it comes to semantic technology and ElasticSearch. While speed isn’t as important to an M&A investment bank as it is for a prop-trading shop, when dealing with massive datasets, speed can become an issue. To help improve performance, Lubell is conducting a proof-of-concept using the Julia programming language. The system employs a taxonomy and large topology that was created internally. Julia is an open-source language that is faster at weeding through large datasets to produce results than other research languages like R, Python and Matlab. (See sidebar.)
“Julia is 20 times faster than Python, 100 times faster than R, 93 times faster than Matlab and 1.3 times faster than Fortran,” Lubell says. “What really excites us is that it’s interesting that you can write high-level, scientific and numerical computing but without having to re-translate that. Usually, if you have something in R or Matlab and you want to make it go faster, you have to re-translate it to C++, or some other faster language; with Julia, you don’t—it sits right on top. That will be one of our big investments for the coming year.”
Lubell says what they’ve built at Berkery Noyes has other capital markets applications. He’s a big advocate of semantic technology and believes that it can be used to help trading shops link together comparable securities—i.e., “I’m interested in Company A, which has these characteristics; can you recommend similar securities in this space?”
Additionally, Lubell sees the potential to incorporate machine-learning tools into the platform as a likely next evolution, although they are still in the examination stage. He also believes this could serve as a first step in lessening the firm’s reliance on email. As is true with most Wall Street firms, while fax has largely been conquered, email is still a primary means of communication and serves as a bootleg client management repository. A potential next level—though this is still in the development phase—is creating a customer engagement website that creates something of a community that allows users to interact with the bank’s data more directly.
This could also open up a new avenue of growth for the bank. Right now, Berkery Noyes specializes in M&A advisory for mid-sized deals, although there’s an untapped market of smaller deals available to be examined. Smaller deals—about $5 million or less—are usually not profitable enough to pursue. But with what is being created, a robo-investment bank could make that segment more viable.
Through his DoD experience, Lubell says he learned the importance of creating environments and also that Waterfall development programs can prove difficult to live with—Agile is the way to go. “Waterfall projects in software generally fail. It’s only since the early 2000s that people have really gotten a handle on how to manage software with the Agile manifesto and the idea of doing quick, iterative things,” Lubell says.
Berkery Noyes uses “sprints”—two-week projects to accomplish a desired task, and then takes a step back to evaluate the results. While he learned much working for the government, the DoD wasn’t Lubell’s only influence. For example, while in Spain—playing music, painting, working with stained glass—he never lost touch with technology. While he disengaged from tech for a bit, he never left its orbit and eventually he took on consulting work that would lead him to Berkery Noyes.
“Because technology was moving so fast, you could kind of jump in because everyone was still learning new things all the time,” he says. “There were still mentions of technologies that I had never paid attention to that would’ve been advantageous, but it wasn’t so big because in the technology business you have to learn all the time. You can’t sit on your chops.”
For six years, Lubell explored Barcelona and honed his studio-arts side. As technology and creativity increasingly blend, those skills aren’t quite as divergent as they might have once seemed. Tied together, Keith Lubell’s experiences have crafted a well-rounded technologist, which is reflected perfectly in the platforms he builds.
LUBELL AND JULIA
It was Keith Lubell who first alerted Waters to the potential uses of the Julia programming language. Seven years ago, four individuals came together to create this fast and expressive language to compete with the likes of R, Matlab, Python and about three dozen other dynamic tools. Waters took an inside look at how some finance firms and economists are using Julia, and examined why this year could be a year of significant growth for the upstart.
To read more, click here.
Keith Lubell Fundamental Data
Name: Keith Lubell
Education: B.A. in physics from Middlebury College
Hobbies: Bass playing, 3D integral photography
Top 3 Tech Projects:
• Migrating to NoSQL-Semantic technology to improve business intelligence derived from company data
• Incorporating Julia to improve semantic algorithms
• Building automated dev-ops to improve software quality and speed development
Greatest (Business) Success: Using semantics, data and business intelligence to improve lead generation
Most Significant Influencers: Nicolas Tesla, Buckminster Fuller, Johannes Vermeer
Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.
To access these options, along with all other subscription benefits, please contact firstname.lastname@example.org or view our subscription options here: http://subscriptions.waterstechnology.com/subscribe
You are currently unable to print this content. Please contact email@example.com to find out more.
You are currently unable to copy this content. Please contact firstname.lastname@example.org to find out more.
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Printing this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email email@example.com
Copyright Infopro Digital Limited. All rights reserved.
You may share this content using our article tools. Copying this content is for the sole use of the Authorised User (named subscriber), as outlined in our terms and conditions - https://www.infopro-insight.com/terms-conditions/insight-subscriptions/
If you would like to purchase additional rights please email firstname.lastname@example.org
Voice of the CTO: Enterprise architecture aims to tie together a bank’s overall infrastructure, systems and workflows. While easily said, bank technologists explain why it’s harder in practice.
A summary of some of the latest financial technology news.
John Ameriks, head of Vanguard’s Quantitative Equity Group, explains the rationale behind dataset selection and how the group has been using machine learning.
The London-based market data provider is aiming to provide higher quality data than its new competitors, but it faces an uphill battle.
The exchange group has migrated its core data to the cloud as part of its 10-year partnership with Google Cloud, and is now looking ahead to further migrations and leveraging the cloud to power new data products.
Voice of the CTO: As banks strive to take advantage of cutting-edge tools, the basics of proper data management and governance are too often overlooked. Banks are learning that a CDO can help gain favor with the C-suite.
The Data Excellence Program—the first to recognize data management at an enterprise level, rather than being an accreditation program for individuals—will help firms solidify data management as a program rather than a series of projects.
There’s a novel kind of data tsunami on the horizon: the volume of new datasets, and trying to discern what’s out there—and how much of it is useful.
- Dora technical standards shoot for break in the clouds
- Tick History – Query: Looking back to the future
- AFTAs 2023: All the winners and why they won
- Breaking down the walls: Enterprise architecture gets its day in the sun
- Interoperability is not AI
- AFTAs 2023 winner's interview: Northern Trust