At Mutt Data we’re always looking for new creative, passionate Data Nerds who love applying machine-learning to solve tough business challenges. In a recent blogpost we covered 5 Reasons Mutt Data Is A Great Place To Work and gave you a sneak peek at our hiring process. A big part of that process is interviews so we thought we’d shake things up a little 💃, have some fun, and imagine what interviewing Mutt Data for a job at well… Mutt Data, might look like.
Without further ado, Mutt Data, we’re ready for you:
We officially started our Machine Learning (ML) consultancy business in late 2018 but our passion for data science and machine learning dates far earlier. We’ve been implementing ML systems for more than seven years and working on big data infrastructures for over ten. Presently, We’ve successfully deployed more than fifty global projects for different customers.
Currently we’re a team of +40 Data Nerds working on the forefront of Machine Learning, AI, and Big Data. We solve challenging problems using complex tooling such as object tracking in video, machine learning over graphs, Adtech Optimization and reinforcement learning among others.
Our state of the art solutions are the product of innovation and continuous learning. We’re experts in a variety of modern cloud and new school data engineering practices and use the latest tools and technologies, such as: Athena, Airflow, Spark, Glue, Pytorch, Kubernetes, Sagemaker and MLFlow.
That’s really where the team and it’s processes come in. Working on any project at mutt data means being surrounded by, working with and learning from a team of multi-disciplinary experts who are always willing to help. There’s no limit to what can be learnt.
As to work processes, we’re strong believers in collaborative, async processes and scrum-like, agile work methodologies. The team is remote-first and result-driven, our culture is based on trust and ownership.
We work with local companies such as Naranja X and ANK as well as international clients such as Mercado Libre and Claro.
In three years time we envision ourselves being a top three machine learning/data science company in South America, working with more than 80 mutters’ and reaching out to worldwide markets.
Even further down the line, we’d like to be a leading company in the LATAM market working with more than 300 mutters’ and offering finished core products.
Our immediate goal is to work on 15 new projects, double our earnings as well as our team count in 2021; we’re already well on our way to achieving said goal.
What’s most interesting is managing these milestones whilst remaining current, continuing to research and use cutting edge tools and tech and remain efficient and agile as our team continues to grow.
In the short term, we’re looking to hire Data Developers, Machine Learning Engineers and Data Architects. More on those positions 🔎here🔎
We recently wrote a blogpost on 5 Things that make Mutt Data company a great place to work at, to summarize though, we look for a positive work environment, the opportunity to innovate and continuously learn, the chance to work in a team-driven business with a multi-disciplinary approach to work, a company with engaging work on end-to-end solutions and finally a remote-first culture.
The ideal scenario is working with experts, learning from and with them no matter what project we’re in. A great team is the best way to learn. A company with clear goals and a set of values visible in everyday work is fundamental.
There isn’t one specific type of project because the potential reach of machine learning in different industries and business departments is almost limitless. However, the most gratifying projects tend to be the ones where we can work on end-to-end solutions with the customer.
One can really add value to a business by getting to know the customer, taking the time to identify opportunities, assess skills, tools and tech in that company. This allows for proper identification of issues, thought-out planning and correct data prep. The objective is to make a custom solution that achieves the client’s goals efficiently and works seamlessly with their infrastructure and product.
Some products that stand out are:
Forecasting & Supply Optimization: In this project a B2B food delivery startup needed to find a way to efficiently assign schedules to delivery staff and reduce costs. We built a ML system that uses historical operational data to accurately forecast real-time demand, minimizing the number of people and costs.
Real-time bidding (RTB): On this occasion a company needed a system that would efficiently price each ad auction while processing thousands of auctions every second with response times under 20 milliseconds. Using software engineering, cloud infrastructure and business know-how we built a stable and efficient infrastructure with a new pricing model for simplified optimization and combined it with a ML system that combined different real time estimations to accurately price auctions.
Marketing Optimization for Ecommerce: Our customer, one of the biggest spenders in Google SSC, DSA and Adwords, worked with multiple marketing teams to operate their campaigns. The lack of cross-team best practices and strategies was leading to suboptimal local solutions. We worked together to develop a fully automated ML system that explored and exploited all available data to achieve robust optimization across all campaigns. Daily predictions and constraints and cross-campaign information are used to suggest several budget and campaign configurations to deliver higher revenue. We also implemented scheduled updates to facilitate decision making and communication.
The foundation for any project is transparency and active communication; keeping people in the loop. However, each solution is unique and custom made so it’s key to have a process, the one thing you can standardize. Our process looks something like this:
We meet with our clients, map out objectives, constraints, possible tradeoffs and priorities. Not all companies know exactly what they need from the get-go. By discussing these variables we can design a custom solution understanding what we need to control and what we can and cannot adjust moving forward.
Through standardized engineering and design questions and machine learning canvases we get to know their current infrastructure and illustrate our vision in an easy way communicating what we want to achieve. The final result is a product brief, defined metrics, a design document, an architectural diagram and a roadmap with milestones.
We kickoff this phase by reviewing our objectives and roadmap with your team. We make sure progress is measurable so that we can have a clear definition of done and keep the project on track. We then select technical tools and set up Slack Channels, Git repos, Gdrives and other tools to keep work transparent and collaborative.
We build the infrastructure and systems using a scrum development methodology. Our focus goes into internal and external communication. We keep relevant people in the loop, aside from the channels we set up , and a heavy use of slack we organize weekly client meetings, key stakeholder meetings and internal meetings to review progress.
While developing the solution we also apply time-proven Software Engineering practices such as breaking down tasks into issues that can be solved in limited scope pull-requests that are easy to review, setting up clear production and development environments, keeping documentation updated and setting up CI/CD systems to ensure a baseline code quality and automate daily deployments.
Putting a data product into production requires our team to be in sync, everyone’s skill sets are necessary. We follow guidelines, lookout for risks and keep the client teams informed as we proceed. The main objective is for deployment to be a non-event.
Once we deploy we go onto the hardening phase, we take some weeks to test the system in the production environment while it’s being used by the final users.
We make sure your team understands the system as a whole, how it works, how to operate it and, if wanted, how to enhance it. We also schedule internal Mutt talks to share learnings. Without exception, there are always learnings, sometimes technical and specific to Data Science and Machine Learning, or Software Engineering practices or sometimes relating to the business or project management approach. All of this is documented and used in future projects and mutter onboardings.
We hope you’ve found this post useful, and at least mildly entertaining. If you like what you’ve read so far, got some mad dev skills and like applying machine learning to solve tough business challenges, send us your resume!
PS: It was us all along 😄