Want to Ramp Up Your ML Team? Hold a Hackathon

Fortune 500 companies are using hackathons for everything from product prototyping to talent retention and recruitment. And they’re a great way to build up your machine learning team as employees collaborate and quickly iterate on creative ideas. 

Internal hackathons are especially common for tech companies, because they allow employees to work together across departments to come up with innovative ideas and to build stronger team relationships. External hackathons, on the other hand, provide an opportunity to find new professionals in the ML field to build out your own team.

With the pandemic subsiding, now is a great time to hold a hackathon. After spending so much time working remotely, your ML team will be able to reconnect and start new relationships with teammates who joined during lockdown. It also adds excitement to the typical workweek, offering employees the chance to innovate outside the confines of standard work responsibilities. 

By hosting a hackathon, you can take advantage of recent advances in ML/AI tooling, which make it easier than ever to get started with ML as you execute on your ML strategy and ride the wave of most recent ML innovations. Here’s a deeper look at the benefits of hackathons, and how to get started.

The Benefits of Holding a Hackathon

Hosting a hackathon provides an abundance of organizational benefits. These events can

1. increase innovation,
2. strengthen connections across teams, and
3. provide an opportunity to recruit new candidates.

Spur Innovation

Hackathons provide developers with the opportunity to dive into new ideas without worrying about the responsibilities of their day-to-day work. Developers can explore creative ideas with little to no risk, often leading to the generation of novel ideas that might not be considered in a typical work environment. They also can try out new frameworks and tools, so developers can test out disruptive technologies that they might not otherwise have the chance to explore.

Recent advances in AI/ML tooling have made it easier to execute innovative ML projects, especially within the time constraints of a hackathon. There are tools for speeding up the process of collecting and labeling data, as well as for implementing an initial ML prototype. You can find many publicly available datasets across a wide range of domains using resources such as Google Dataset Search and Kaggle. For example, you can find datasets to explore trends in COVID-19 case numbers, to classify emotions in speech audio, and to perform sentiment analysis on tweets.

If you’d rather work with your own data, automated tools can speed up the labeling process, while cloud platforms such as Azure Machine Learning and Google Cloud AutoML help construct an initial ML pipeline. These tools give developers the flexibility to explore more complex problems within the scope of a hackathon, since the developers don’t need to implement everything from scratch. A hackathon also gives participants a chance to try out these tools and see how they can be incorporated into various company projects.

Because hackathons provide a dedicated period to focus on a single project, they are also conducive to fast iteration, allowing developers to create prototypes within a short window of time. These prototypes are often the launching pads for new ideas and products. 

Hackathons also provide a quick way to get user feedback based on a product prototype. The judging panel, typically made up of leaders within the company, gets to try out and weigh in on the different prototypes, and the other participants in the hackathon can weigh in as well.

Build Inter-Team Unity

Hackathons provide an opportunity for cross-department collaboration that isn’t always possible in a constrained work environment. Employees can form teams with members of different departments, opening up opportunities for creative development and encouraging stronger relationships across the organization.

Hackathons also provide a fun environment for employees to work together on less traditional projects. Holding a hackathon for your ML team can be a great way to build a stronger sense of camaraderie among members of the team.

Recruit Candidates

External hackathons provide a great way to find new professionals in the ML field you might want to hire. By hosting an external hackathon, you can easily see whether candidates have a practical knowledge of ML tools and techniques. Some companies organize hackathons specifically as part of their candidate selection process.

Host an internal event to promote team building. Plan an external event to boost recruiting efforts.

How to Set Up a Hackathon 

Once you’ve decided to hold a hackathon, you need to start planning. Even nontechnical people can participate in the project management, planning, and design functions. Cross-department collaborations are a great way to unlock new innovations from a nontechnical perspective. 

You may want to start out by deciding a theme for your hackathon based on your industry. For example, you could focus your hackathon on developing customer-facing products, building internal tools, or creating applications that have a social impact.

Who should organize a hackathon?

The people responsible for planning the hackathon can vary from one organization to another. In smaller companies, hackathons are typically planned internally by engineers who take the initiative and push the plan forward. Larger companies, on the other hand, might have an event coordinator plan the event. And organizations that don’t have the bandwidth to plan their own hackathons can use external event planners to set up an event, since no ML background is needed to organize it.

Logistics and catering

Most of the planning for a hackathon involves figuring out logistics. You’ll need to select a date and reserve a location for the event. Any remote employees will need to be brought on-site, while keeping in mind any COVID-19 policies. 

You’ll also want to organize catering and set up fun events to keep the energy high during the hackathon. Options include poker tournaments, scavenger hunts, and other team-building activities. Finally, you’ll need to assemble a judging panel, typically composed of company leaders, and come up with the categories to judge, such as “Most Innovative,” “Best In Show,” and “Most Immediately Effective,” as well as the prizes associated with each.

Other planning tips

The rest of the planning process involves setting things up in such a way that the hackathon runs smoothly. Beforehand, you’ll need to host a planning session to facilitate idea generation and team matching. You can schedule virtual sessions to pitch ideas, or just have employees add ideas and team members to a spreadsheet. 

Once you’ve set up the teams, they can register for a time slot to present to the judging panel. Then, on the final day of the hackathon, you judge the projects, announce the winners, and award the prizes.

Next Steps

Hosting a hackathon provides great opportunities for both recruitment and team building and allows participants to innovate on exciting, new projects. With recent innovations in AI/ML tooling, it’s easier than ever to get started with ML projects. 

Before setting up your own hackathon, first decide whether to host an internal or external event. Which way to go depends on your ultimate goal. If it is to promote team building and prototype development, you want an internal event. But if you are looking to recruit potential team candidates, host an external or hybrid event. Once you’ve made that decision, you can start planning, setting your organization on the path to ramp up your ML team.

Bihan Jiang helped to coordinate and judge at Treehacks, Stanford University’s annual hackathon, as part of the Scale AI's sponsorship program.

Learn More

1. 7 Elements of a Well-Thought-Out AI Strategy
2. How to Translate Data Science to Business Value: A Blueprint