Lotame's Chief Data Scientist, Omar Abdala, has been awarded 4 under 40 award by the…
This week’s Under the Hood features Omar Abdala, Lotame’s Chief Data Scientist. While Omar isn’t turning data into actionable insights and building out our Data Science team here at Lotame, you will probably find him supporting his kids at various sporting events! Interested in learning about his fascination with autocross and track racing? Read on to learn more!
A “Chief Data Scientist” is, firstly, a “Data Scientist”, so let’s understand what that is. As computing power and data storage have gotten cheaper, and big data technologies have allowed many enterprises to store all sorts of data, there’s been a peaked increase in the need for technologists who are able to utilize these reams of data and turn them into actionable insight. We have come to call a person who is responsible for that a “Data Scientist”. The job title “Data Scientist” didn’t exist when I entered the workplace and as with any new nomenclature, definitions vary.
My personal definition has been that a data scientist has more analyst skills than a typical engineer (i.e. has knowledge of Statistics, Optimization, Model Solving, etc) but also is more of an engineer than most pure analysts (i.e. has knowledge of programming in general, and specific knowledge of a few Big Data platforms sufficient for implementing practical solutions). As the field of Big Data didn’t exist as such while I was studying, I came upon these skills almost by chance. First, through an undergrad degree that had a steady dose of Software Engineering, then later graduate work in Statistical Signal Processing and self taught practical distributed computing when I needed to use big data platforms in my early work in Marketing and AdTech. I think this kind of serendipitous path to becoming a Data Scientist was typical as I became one. Now, as for a “Chief” data scientist, I think there are two important additional pieces: 1) To be able to distill customers needs into product direction for analytics and related tools and 2) To have sufficient expertise to be able to advise other Data Scientists on their work and reduce the round trip times between ideation, successful prototyping, and product design.
There are plenty of challenges in transitioning from a company with its own customer set, priorities and roadmap to a company with different ones, some of them unknown and unexpressed for months as strategy is defined. Managing people through such a transition is by far harder than managing product alignment. One of the key things that I learned about the culture at Lotame, appreciated, and carried through my relationships was complete honesty and transparency about what we’re doing with our products, what that means for each individual, what opportunities we have within Lotame and even sometimes, what better opportunities might exist outside Lotame. The goal is to give each person the information that would allow each of them to make decisions that made the most sense for themselves and their families. When we say “Lotame is Family”, that translates into something quite real about the way we interact. In a self centered culture that views each individual in terms of “what can I gain from him/her?”, this type of genuineness is unique and can be described as our open “secret”.
I joined the AdTech industry in 2009 when I went to work for Quattro Wireless, one of the first mobile ad networks (Quattro was doing mobile ads before the iPhone — like the Palm phones and Windows Phones where the OS looked more like Windows 95 than iOS) so I’ve been at this for a while. The AdTech industry has experienced some change but we shouldn’t forget the two industries that collided to form our current culture. On one hand, we have advertising, an industry that is at its core about showmanship where overselling does seep from the ads we place to the services we as an industry sell in order to place them. On the other hand, we have the tech and startup culture dominated by exhortations to “fake it till you make it”. As technologists in ads, there are strong incentives in our industry to veer from: 1) our core scientific principles. We are at our hearts skeptics; we want to measure everything and challenge assumptions. AdTech can eat away at these inclinations in the pursuit of “the show” and 2) our core moral principles. There’s always a dollar to be made from ad optimization of predatory loans to the underprivileged or for profit universities to prospective students who don’t know better than to take huge loans. In the long term, our customers will judge us (this makes (1) a losing strategy) and society as a whole will judge us (this makes (2) a losing strategy). So the big lesson is: despite all of the pressures to do otherwise, make sure to place yourself in a team that supports you to stick to your principles.
I think a good manager is someone who can motivate and position a person who would otherwise be a mediocre contributor to be a high quality contributor. Under this definition, I don’t consider myself a good manager. The reason is I only hire truly excellent people. In all the years, over the 4 companies I’ve done hiring, I have never once regretted a hire I’ve hand picked. Some may say I don’t hire quickly enough, but this is the way I like it. I’ve been blessed to never have to build out a very large team from scratch in an instant. Therefore, I hire only exceptionally qualified and highly intelligent people and then I don’t interfere. Where I do try to exert myself is in impactful and meaningful goal setting, technically or organizationally.
I’m going to talk about the Data space (as opposed to DMP, Marketing Research, TV analytics, etc which are all spaces that will change and Lotame touches). The data products we’ve been delivering at Lotame are a reflection of where I believe the industry is moving. Data is now a mature industry. We’re no longer at a stage where targeted campaigns can be sold on vague promises that “behaviorally targeted campaigns are more performant” or “reaching the right user limits ad fatigue” Each datapoint must prove its performance value, accuracy, or both.
As a large 3rd party data provider, I believe we’re in the first phase which is creating parallel datasets where we guarantee some metric the data buyer cares about. Phase 2 may be altering the pricing structure on data based on performance. Why should highly accurate Precision Data segments be sold for the same price as data which we can prove is not much better than random? The next phase after that is likely to be a consolidation/reduction of the volume and number of data providers in the 3rd Party data marketplace. Only those who can do end to end modeling and provide provable bounds on campaign outcomes with their data will be able to survive.
Much of my non-work time is occupied by kids’ sports and other family stuff (like everyone else). But one unique thing I’ve gotten into recently and I find fascinating is autocross and track racing. I’m amazed by idea that you can make a minuscule change to the exact moment you brake at a corner, or when you start your turn in that can impact your ability to apex correctly and exit a corner setting yourself up angled correctly and moving faster that can shave a lot from the lap time. The combination of: the math/physics that lie at the heart of it, the technical/execution excellence required, and obviously, the thrill of zipping faster than you should and occasionally losing traction is awesome. I don’t know how or why I lived for so many years without experiencing it.
I think people learn a lot from their first working experiences. My first job was fixing Electrolux Vacuum Cleaners in East Brunswick, NJ. Mostly Diplomats and Ambassadors (https://evacuumstore.com/c-1908-electrolux-history.aspx) We sometimes think of today’s Dyson vacuum cleaners as expensive, but these things were *seriously* pricey. $800 in 1995 money is ~$1325 in today’s money and at that price they certainly were worth fixing. I got the job by walking in as a (maybe 15 year old) and telling them I’d do anything. I expressed an interest in fixing, gave them my word that I was handy with tools and that was all they needed. I remember being obsessive about details like screws being stripped inside the machine, or rust on the motor housing. I wouldn’t want to let the screw go without loctite (or whatever compound we had) and wouldn’t stop until the rust on the housing was scrubbed fully clean. My dad’s voice was in my ear telling me that only people who perfect the details can expect to be the best. I knew I was not going to fix vacuum cleaners forever. In fact I was still fixing them the summer I was preparing to go off to MIT. I knew that no vacuum owners were ever going to open the case and look at the work I’d done. I also knew that if I did a half job, I could clear more vacuum cleaners and make more money. Yet I still wanted the satisfaction of knowing that the job was done right. Call it my first rejection of the dominant purely capitalistic paradigm. I’d like to think that some of that young man’s vigor for work remains today. If I could talk to him, I’d tell him not to change a thing.