My guide to Surviving a Hackathon (with RPDR)

#hackathon #ai #machinelearning

In April, I have attended the Good AIdea hackathon in Belfast, alongside three of my friends. The aim of this hackathon was to design a solution using AI.We went as a team, learned a lot and actually had a demo to show by the end of it.
It was their first time, mine second and I thought I would share my experience with you all, so hopefully you, can attend if you ever wish to.

Hack-a-what?

A hackathon is an event during which a team of people, "hack" a solution together based on a given challenge or theme in a given time. Prototype-for-your-LIFE!

Whilst they can seem scary, or you think you will never be able to make anything in the timeframe, they are a great opportunity to learn. You will also be able to meet fellow developers, expand your field of knowledge, collaborate and meet some recruiters.

And even if you do not attend to code a project, many of the hackathons offer workshops and talks to upskill attendees. There is something for everyone!

What you will need on the day

Usually, you will require a laptop. If you don't have one you can usually pair with someone or ask the organisers if they can source one for you, for the duration of the hackathon.

Contrary to popular belief, you don't need to go with a team. Many people come along and join a team during the event.

Other than that, bring a reusable cup, as coffee is usually provided. We love an environmental queen.

Traveling to a hackathon

We had to travel from Scotland to NI which was a quick flight. We have arrived the night before, had a takeout, chatted for a wee bit and then went to bed.

Hackathons usually run during the night, and many people choose to sleep in the venue.

I am not that person.

Before the event

Set up your environment as much as you can before attending. This includes your AWS/Azure account, code editor, version control etc. I think this is one of the most important preparation steps. You can't always rely on Wi-fi ass there are probably other 200 people on the free network all trying to download Android Studio in a mild panic.

Remember, part of being prepared is knowing there is usually a shop next to you with free, underutilised wifi.
If you know the themes, have some project ideas - they just need to be little seeds, no need for full plants.
Understand what you want to achieve and let that drive you. Personally, I knew that at the end of my hackathon experience, I wanted to have a demo to show, with an AI component that filled the brief. When scoping the project I kept concentrating on that, which helped me to narrow down my work.
Remember, sleep is important.
HYDRATE!

Start your engines, and may the best project, WIN!

So on the day, we were given 3 major challenges to pick from :

Fake review recognition
Improvement of current design/software of hearing aids, esp in busy environments
Build something "for good" which had an AI component

We have chosen to go for option 3, as we weren't trained in AI, and as mentioned before, we wanted to have something working by the end of the hackathon.

Our strategy was to come up with some ideas, narrow them down and then research them for an hour, to see how realistic they were. Obviously given the time constraints, good research and deep team understanding of the problem was worth more than the disappearing mountains of Monster cans combined.

What did we build?

Our idea was to build an app which would help users to understand if a product was vegan or not. People are becoming more aware of their food choices, and there is not much regulation for displaying animal product components in food. In addition to that, it can be super time-consuming to read everything.

Introducing : VegAI

We made an android app, which used the Google Lens API to determine a list of ingredients from the product. We have then had our own serverless amazon lambda function which checked the ingredients against a blacklist.

We focused on 2 products - Monster can which was vegan and KitKat which was not vegan.

The way we divided the team was by asking who wanted to do what. I was keen to do AWS whilst others wanted to try mobile development.

Initially, we wanted to have a model which would recognise the brand and do some other clever things but Google Lens has failed us. We could not consistently determine the product and so we have focused on reading the ingredients list instead.

Before this none of us have ever written a lambda function, used android studio or even heard of dynamoDb. But in 24h we had put together a small architecture.

What did I get out of it?

First of all, it was a great challenge - I have never used AWS and wow, it had a massive learning curve.

Apart from that, it was great fun, it was a really creative challenge and I spent time with my friends. Woop! And if people are not impressed then :

My top 3 Tips for the next hackathon

I do think that the initial research and try out was incredibly useful. I think the time we spent researching, we 10 x saved in development, especially if we were to pick a project we couldn't complete.
Talk it out - make sure your team is on the same page. Communication is key, in any successful team, and hackathons are not different. Teamwork makes the dream work!
Use a pre-existing data set - esp. in an AI hackathon challenge, try and find a ready dataset as otherwise, you will spend more time collecting data rather than developing the product. This probably represents the real world accurately, but at a hackathon you just want to get stuck in.

Will I do it again?

Hack-a-problems:

As much as an amazing event it was, from looking online and reading some other blogs hackathons have a bit of a bad rep, and in general, there are things which can be improved.

This is what I understood the problem across hackathons is that teams consist of students who, don't need to work weekends or took the time off to do so. I appreciate the time commitment does narrow the type of attendees quite significantly. Some steps to fix that:

Allow and facilitate for remote work teams/members.
Allow for part-time participants, explicitly advertise this! Also, advertise your workshops.
Speak to your (potential) attendees to understand what barriers they face to attending your event. For example, childcare or your venue accessibility constraints, not knowing anyone etc.
reach out to people who are not at Uni. Many local meetups would be happy to spread the word to people not at universities such as people new to tech, boot campers, and self-learners!
have a code of conduct to make sure everyone feels safe and welcome. This means having a written conduct but also
- outlining clearly how someone can raise the issue to you
- having a clear process for raised issues
- having a dedicated and trained team to deal with issues

From what I can see, not many career changers attend hackathons - not only for the reasons above + many hackathons seem like they are directed at CS traditional students only. Bleh! Why put unnecessary constraints on yourself? Reach out to people outside your sphere!

And on another totally different note:

I really enjoy the art of drag, and I would like to encourage you to support your local drag queens. They do amazing work, for a 1/100th of the budget when compared to RPDR. In Glasgow, we have DragOpticon, a sell-out drag show, first Friday of every month!