CTO Sessions: Alexander Hudek, Kira Systems

Which emerging technology are you most excited about the prospect of? "On the AI front, I think differential privacy is going to play a bigger role across the industry in the coming years."

IDGConnect_ctosessions_suppliedart_alexanderhudekkirasystems_1200x800
Kira Systems

Name: Alexander Hudek

Company: Kira Systems

Job title: CTO & Co-Founder

Date started current role: May 2011

Location: Toronto, ON

Alexander Hudek is Kira Systems’ CTO and Co-Founder, and leads the company’s products, technology and research departments. He holds a Ph.D and M.Math degrees in Computer Science from the University of Waterloo, and a B.Sc. from the University of Toronto in Physics and Computer Science. His past research in the field of bioinformatics focused on finding similarities between DNA sequences. He was heavily involved with the human genome project, helping create the first full draft of human chromosome 7.

What was your first job? I worked in a research lab that was part of the human genome project at The Hospital for Sick Children (SickKids). One of the most notable projects I helped complete was creating the first full draft of human chromosome 7. I also created an analysis framework for finding new genes that was used for many years.

Did you always want to work in IT/technology? Yes. I’ve always been interested in computer science and began programming computers at age eight. In undergrad I took courses in algorithms for planning and logic, machine learning and AI, numerical computing, and other topics. My interest in machine learning grew more specifically during my PhD at the University of Waterloo. There, I used machine learning methods to study DNA. Afterwards, I dove more deeply into formal logics as part of my postdoctoral research. Logic and reasoning is in some ways the “other side” of the coin in approaches to AI and I felt it important to know more about it.

What was your education? Do you hold any certifications? What are they? I have a Ph.D and M.Math degree in Computer Science from the University of Waterloo, and a B.Sc. from the University of Toronto in Physics and Computer Science.

Explain your career path. Did you take any detours? If so, discuss. I had recently attained a Computer Science Ph.D. from the University of Waterloo, and was completing post-doctoral research when my co-founder, Noah Waisberg, reached out with a business idea in 2011. He thought it might be possible to build software to help lawyers find and extract information from contracts, but needed a technical partner to help. The problem was interesting to me because it involved natural language. I hadn’t done anything previously in natural language, and it had always interested me. There is something about human language that really makes it feel close to human cognition. I joined him, and by 2013, two and a half years later, the software was able to help customers do contract review in 20% to 90% less time compared to manual review.

What type of CTO are you? Truth be told, these days I do more product and business thinking than I do technology. I still keep on top of tech by reading papers and playing around with technology in my personal time, but for the day to day I delegate to leaders that report to me, providing only higher level guidance. If you want people to truly take responsibility for their work, you need to give the space and trust to make decisions without you always watching over their shoulders. If you can’t do this, you don't have the right people working for you.

Which emerging technology are you most excited about the prospect of? On the AI front, I think differential privacy is going to play a bigger role across the industry in the coming years. Privacy is increasingly an important topic to everyone, and with the number of security breaches happening it’s clear that security practices alone are not enough to truly protect the privacy of data. Deep learning is obviously a very promising technology as well, and it’s exciting to watch it quickly evolve.

Outside of AI, I’ve found the Go programming language to be making a solid impact on the industry. Although many people don’t find it an “exciting” language, its boring simplicity and enforced coding standards are a huge boon for large teams. Clever code isn’t good when you need to optimise for others understanding it. The performance and deployability of Go code is also impressive. Computing is so cheap these days that often people underestimate the value of using efficient languages when it comes to CPU and memory usage.

Are there any technologies which you think are overhyped? Why? Although this answer is boringly contrarian, deep learning is definitely over hyped. In our own studies we’ve only managed to get a few percentage points of improvement in accuracy on the problems we care about using deep learning. While this is great from a pushing the boundaries point of view, it’s not worth the trade off of 1000x slower training and prediction algorithms. At scale hardware costs become real hurdles. I do think deep learning is the most promising avenue for language understanding that we have today, but it’s not yet as big of a leap in this domain as much of the media suggests. This is often the case with science reporting sadly.

Another technology that I feel is overhyped is microservices. This is an architectural pattern for building distributed systems. Although it can be useful in many circumstances, it’s not a magic bullet that is appropriate for all situations. Unfortunately, that’s often how it’s positioned. In reality there are real trade offs and downsides to microservice architectures that are important to understand and weigh. For example, although yes you can often horizontally scale a microservice, you also incur more network overhead. Depending on how “micro” your microservice is, this may or may not actually give you improved performance. It all depends on the application.

What is one unique initiative that you’ve employed over the last 12 months that you’re really proud of? Kira is AI-powered contract review software that enables teams to truly know what’s in their contracts and documents. It comes with 1,000+ pre-built machine learning models for the most common contract review tasks such as due diligence, deal points tracking, commercial contract analysis, lease abstraction, ISDA schedule reviews, and more. Our customers can also build their own machine learning models with a capability called Quick Study.

In September 2020, we took Kira to the next level of contract analysis technology by launching Answers & Insights. This new capability goes beyond identifying and extracting provisions, clauses, and data points, and labels them based on the meaning behind the text. This provides decision-makers in firms and organisations with the answers to their most pressing questions. Questions like: “Is LIBOR or Eurocurrency referenced in the agreement?”, “Are there environmental indemnifications in the lease?”, or “Does the lease require the tenant to obtain business interruption insurance?” Decision-makers in firms and organisations will gain a deeper understanding of what their data means, helping them make faster, smarter decisions or recommendations for their businesses and clients.

Even better, Answers & Insights is built into our Quick Study platform, letting anyone teach the system to answer new questions without the need to know how to program or have deep machine learning knowledge. It does all this while protecting the data you use to teach the systems using the strongest privacy preservation techniques in the industry.

Are you leading a digital transformation? If so, does it emphasise customer experience and revenue growth or operational efficiency? If both, how do you balance the two? I’ve been pushing our company to use our AI technology in novel ways internally. It’s definitely hard to find the time to do this, especially when the application is not something that has been done before. Ultimately we are doing this to be able to offer our customers better experiences, and also to test new applications and products before launching them. Improving operational efficiency is both a bonus but also a must. If we can’t get a good ROI on an application of technology ourselves, it would be dishonest to subsequently sell it to others! Overall, we lean towards customer experience and future revenue with these projects. I can’t tell you more today, but perhaps you’ll see some of these use cases in our future products.

What is the biggest issue that you’re helping customers with at the moment? 80% of data is unstructured and even though contract analysis technology transforms the way businesses extract information from within documents, this only scratches the surface. With Answers & Insights, users can quickly answer questions across their entire set of contracts. This will allow businesses to react far more quickly than they do now, and understand all their contractual relationships in a way that is unthinkable today.

How do you align your technology use to meet business goals? To me, it actually starts with customers and the problems they need to solve. From that you can choose business goals, and the product design and technical goals need to follow. If technology and business goals are going in two different directions, you’ve already gone off the rails. That said, your existing technology can help inform which customer problems you want to prioritise from a business standpoint. It often makes sense to expand into problem spaces where you already have good technology alignment. But at the end of the day the customer's problems have to come first.

Do you have any trouble matching product/service strategy with tech strategy? While we strive to have our technology strategy follow our product strategy, we’ve definitely had times where technology choices have impacted things we’d like to do on the product. It’s not possible to see all future needs when making technology choices, so mismatches are inevitable. The key is to build a culture of fast iteration and change into your organisation. When people get hung up on how things were done before, and are not used to a faster pace of change in processes and architectures, then technology-product mismatches can really slow you down. Past challenges here have really driven this perspective home for me. I always knew the value of iteration, but still ended up undervaluing it in the early years.

What makes an effective tech strategy? Innovate only where you must, otherwise use boring technology. Businesses do need unique differentiators, but the majority of what they do is the same. It’s tempting to innovate everywhere, but this will end up creating more technical debt and risk than necessary.  When it comes to innovation, academic literature is a gold mine of good ideas. There are hundreds of startups waiting to exist, that can be based on unsung advances buried in scientific journal articles. My rule of thumb is first look for tried and true technology that will work for a given problem, then look into existing research and academic works, and only then try to invent something truly new.

Many people confuse innovation with invention. Innovation is about applying technology in new ways, while invention is creating truly new technologies. The latter is far more expensive and risky, and most of what you see in industry is actually innovation being presented as invention. As a leader you need to see through this.

What predictions do you have for the role of the CTO in the future? An effective CTO needs to understand the non-technical aspects of the business and how technology affects them. If I had to guess, the CTO role will become more and more involved in product and product design, as it’s often hard to divide these disciplines. A CTO that additionally understands sales and marketing can be especially impactful. You need the deep technical expertise, but also need good breadth.

What has been your greatest career achievement? In 2014, I built a crude version of a long-desired feature (Quick Study) that allowed a person to teach the software to find new concepts. This was a huge milestone for the company because it allowed anyone to teach the system without feeling the need for a technical expert at their side. Clients could now highlight and tag provisions in a document, press a button, and it would learn what to look for. This, plus a market that was getting more and more focused on efficient legal work, ignited the sales of Kira.

Looking back with 20:20 hindsight, what would you have done differently? I would have taken leadership coaching and read up on evidence driving management far earlier. Also, in the early days of Kira I was definitely very guilty of trying to be innovative everywhere. We succeeded anyways, but paid a strong technical debt price, both in terms of being able to hire people with niche skill sets and in terms of suffering immature frameworks.

1 2 Page 1
Page 1 of 2