Machine learning is the driving force behind many of the world’s most innovative companies, making it one of the tech industry’s most economically robust technologies.
According to Tractica, revenue among global AI software companies is expected to reach $126 billion by 2025. And with so much promise being placed on the future of machine learning, it’s no surprise that many companies are harnessing its power to transform their own products in many different — and somewhat surprising — ways.
For some companies, machine learning holds the power to redefine the customer experience, which in turn boosts engagement and acquisition and increases revenue. At other organizations, AI holds the potential to automate processes internally, making it easier for teams to tackle problems and focus on larger goals.
Regardless of how companies use machine learning, it’s clear AI unlocks endless possibilities, allowing companies to spark innovation across a wide range of sectors.
Built In LA checked in with three local tech leaders to learn about the surprising ways they’re using machine learning.
What they do: Beyond Limits develops cognitive computing technology that mimics human thought processes to aid decision-making.
What’s a surprising or interesting way your company is using machine learning?
Beyond Limits is utilizing machine learning in a lot of different ways to tackle some of the hardest problems in the oil and gas industry. The data scientists and AI engineers at Beyond Limits all come from different backgrounds, which gives us the chance to employ a multitude of techniques unique to our experiences in machine learning and statistics when addressing problems. The project I’m working currently on, Cognitive AI for Formulation solution, uses a combination of machine learning, optimization and mathematical modeling to predict certain formulation values in a system.
At Beyond Limits, we integrate machine learning into larger, more extensible workflows. We use it as a tool, although it’s not the only tool: Machine learning allows our products to handle a multitude of situations and adapt more easily to changing dynamics.
What impact has machine learning had on your business, product or the customer experience?
Machine learning has significantly affected the way we build our products and has improved our clients’ workflows and production pipelines. Using machine learning, we’re able to automate certain processes like never before. The project I’m working on would typically require an engineer to spend hours or days experimentally deriving certain values before spending additional time finding potential results to be verified. Our work automates this process using machine learning to derive those values and test and verify those results, which ultimately speeds up entire formulation processes.
Machine learning allows us to examine and extract certain relationships that might otherwise be hard to see in the data, which can be very helpful in finding new and better solutions. This also gives us solutions that are more efficient and reliable while reducing the impact of human error and increasing accuracy. Overall, this has allowed us to tackle problems in processing, refinery operations, lubrication formulation and more. With machine learning, the opportunities are truly endless.
Using machine learning, we’re able to automate certain processes like never before.”
What excites you most about the work you’re doing?
I really like how the work I’m doing is impactful, tackles multiple tough problems and significantly optimizes production pipelines. It’s really exciting to work on pain points in such large industries, retain ownership of my work and play a large role in how some of our systems are built. We are given the flexibility and freedom to use pretty much any algorithm and model we want.
The research and development phase of building out these products is so fun because we get to play around with all these different methods. It shows that we’re always willing to expand our knowledge to use state-of-the-art applications that garner the best and most reliable results. At the end of the day, it’s great to see our products being used by customers and know that we’ve made their processes faster, more reliable and more interpretable. It’s so rewarding to build products with the power to make a difference.
What they do: Bliss Point Media develops tools that enable companies to maximize the effectiveness of their ad campaigns.
What’s a surprising or interesting way your company is using machine learning?
At Bliss Point, we harness machine learning to optimize clients’ TV and streaming video campaigns and ensure the most efficient investment of marketing dollars. Before coming to Bliss Point, I did not fully grasp all the decisions that influence the delivery and performance of streaming and TV advertisement. These optimization problems fit well within a decision science and machine learning framework, offering plenty of opportunities to apply cutting-edge methods to new and challenging questions.
At the same time, we recognize that machine learning, while a powerful workhorse, does not provide a tailor-made solution to every question. I really appreciate that Bliss Point employs the right tool for the right job, whether that be robust causal inference designs for observational data, time-series analysis or experimental methods. A diverse toolkit is necessary to fully unpack clients’ data and improve campaign outcomes.
What impact has machine learning had on your business, product or the customer experience?
Machine learning is integral to optimizing network decisions in creative media. In particular, techniques like regularization help us provide insights for new client campaigns where data is sparse and speed is essential. While our machine learning products most obviously improve client campaigns, they have the knock-on effect of saving time for our analytic and media teams.
By taking the guesswork out of campaign optimization, machine learning allows our other teams to focus their energies and expertise on delivering new insights and better experiences to our clients.
Machine learning allows our other teams to focus their energies and expertise on delivering new insights and better experiences to our clients.”
What excites you most about the work you’re doing?
The most exciting part about working at Bliss Point is being a member of a growing company and the daily variety that entails. The econometrics team itself is small but growing, which allows for plenty of opportunities to engage with multiple parts of the business. On any given day, I may work on optimizing an experimental design, improving a time-series model or developing a machine learning product. There is a lot of room to take ownership and leverage data in new ways to create both client-facing and internal products.
What they do: Thrive Market is a monthly subscription service that delivers healthy and sustainable products to people’s doorsteps.
What’s a surprising or interesting way your company is using machine learning?
A core premise of grocery shopping is that the customer is left to their own devices. You have to find the right aisle and read the labels to select the right product. And if you have special dietary needs, like vegan or gluten-free, it becomes even more problematic. To facilitate that process, Thrive Market has introduced a guided shopping experience where shoppers answer a series of questions about their individual and family needs. They can specify their dietary and ingredient preferences along with social and environmental priorities like BIPOC-owned businesses, ethically-sourced, fair-trade and recycled packaging. These questions are tailored to strongly gauge the preferences of the customer in order to personalize the user experience.
Using the guided shopping questionnaire, product and value preferences get encoded and are then fed into several machine learning models that combine other signals like purchase data. This allows us to drive a personalized user experience across all touchpoints including website, mobile apps, emails, text messages and push notifications. It also grants the customer an experience tailored to their specific needs and interests.
What impact has machine learning had on your business, product or the customer experience?
With the help of machine learning, we have been able to drive efficiency wins in customer acquisition and engagement and increase revenue. New visitors to Thrive Market have appreciated this personalized service, which has significantly improved conversion and has reduced checkout time.
With the help of machine learning, we have been able to drive efficiency wins in customer acquisition and engagement and increase revenue.”
What’s a part of your tech stack that you really enjoy working with, and how are you applying it in your work?
One of the most important skills a data scientist needs to be successful is good judgment, the ability to adapt to the problem at hand and work on an issue that is both challenging and rewarding. My work on guided shopping has given me the opportunity to take what I have learned and apply it to a problem that can help customers have a personalized experience. The tech stack we worked on used Python extensively for building machine learning models. We used various tools and ML modules which include Flask, Redis, SciPy and LightGBM among many others. We also worked on in-house APIs that were deployed for the website and mobile apps to access the recommendations. Lastly, we set up DAGs in Airflow to sync recommendations to different systems.
