Our latest hackathon took on the challenge of machine learning. We showcased talent, intelligence and the dedication of the Applaud team in our second annual two-day event.
Every year, the Applaud team in Hyderabad, India gather to compete in an internal hackathon competition. This year’s theme was machine learning, a topical choice because it is so relevant today in the world of technology and HR.
The brief was to use machine learning to address an HR pain-point. Two of our teams attempted to predict employee attrition using machine learning and real-time sentiment analysis. The third team focused on the hot topic of employee experience and created a system for gathering real employees’ feedback and pulling out the themes for leaders to action using machine learning.
At the end of the two days, each team would demo their hacks to our panelists who would score each team against a set of criteria. No pressure, but the rest of the company were watching to see what happens.
Keep reading to hear from Duncan Casemore, co-founder of Applaud and CTO, about how the teams created solutions and who won Applaud’s 2018 Hackathon.
Day one: Planning and design
We had three teams this year, Alpha, Beta and Gamma, led by our captains Ramesh, Srini and Karthik, respectively. The morning kicked off with a private 30-minute ‘ideas grooming’ session for each team with Hari, Navin and I. Unlike our 2017’s chatbots theme, this year’s topic was broad and completely wide open. I had absolutely no clue what each team would pitch. Each presented some incredibly compelling possibilities and, dare I say it, game-changing potential.
At 20:15, there was no sign of the teams slowing down with pizzas due for delivery. Territories were established within the office; each group huddled around tables with every square inch occupied by laptops. As a stereotypical tribute to hackathons, the screens looked reassuringly technical.
Sleep was not on the agenda.
Day 2: Idea execution and delivery
I arrived at 09:30, and the office was densely populated and buzzing despite acute sleep deprivation. Grandhi had opted to choose the boardroom table for his bed; apparently, its comfort surpassed an adjoining table littered with cables. Vijay didn’t bother with sleep at all. Our contenders’ energy and determination were unaffected.
The work had shifted from planning to execution; every single person had a job to do, and now was the time to deliver. The teams powered through physical exhaustion with relentless, intense focus. Some were furiously deploying the latest code while others were rehearsing their pitch.
At 15.30, all coding had to stop. Each team now had strictly 30 minutes to demo their creations and take questions.
Team Alpha, led by Ramesh, presented a compelling business case for predicting employee attrition. While the basic concept is not new to the marketplace, their approach was more innovative than anything I’ve seen, using both structured and unstructured data. The results boasted 94% predictive attrition accuracy, a working end-to-end solution, a blended mix of machine learning technologies and some very fast and attractive user interface design. Somehow, they found the time between learning all these new technologies and an entirely unfamiliar programming language to produce a video of the solution.
Next up was Team Beta, led by Srini. Coincidentally, Beta had also elected to look at predicting employee attrition but took a wildly different approach. Ingeniously, they had used natural-language processing to use the sentiment, emotions and traits found in ex-employees’ Glassdoor reviews as a prediction model to spot tell-tale signs in employee-authored content such as their performance review documents. It was amazing to see real-time downward trends in joy, happiness and general positivity immediately following submission of a bad performance review. Largely working end-to-end, this demo was packed full of features and had another simple and beautiful user interface. Incredibly, Team Beta also conjured time to produce a video.
Gamma, led by Karthik, took to the stage with a great-looking presentation and a crystal-clear business case. The team focussed on one of my own passions: improving employee experience by listening to the voice of employees and acting to make the workplace an exceptional place. Gamma’s data collection was genius: pop-up pulse surveys with just a single text-box that continually collect employee feedback on a raft of customer-defined topics every single day. And it wasn’t just pulse surveys; similar macro insights for performance reviews were added to the mix. Machine learning collated the individual responses into company-wide insights including strengths, weaknesses and the feedback’s underpinning themes. The features were complete end-to-end and climaxed with a rogue employee writing feedback so bad that an email was immediately sent to HR to trigger an investigation. I wonder if their manager really was an ‘arrogant sod’…
The winners are…
All of us panelists consulted with each other, and 45 minutes, later we were ready to announce the results of the contest. With a unanimous vote, the judges declared Team Alpha the winners! Their solution was so complete, it could almost be dropped into our next major release and had a proposition so strong that I just can’t see anyone refusing it. Kudos to the Captain, Ramesh, and his teammates Akhila, Gayathri, Keertana, Kumar, Nivya, Raj, Sanjana, Sundeep, Vijay and Vinay.
Of course, winning is great, and yet so much of our Hackathons are about having fun, learning new things, working with different people and contributing to Applaud’s future product offering. Each team presented truly astonishing results, and, in less than 48 hours, they assembled and mastered more tech than I’ve seen in a lifetime. Yes, that’s what developers do best; but for all three teams to select the right tech mix to address major HR pain and present such compelling business cases is a testament to the top-tier talent that we’re lucky to have. Every single person contributed, and the teamwork, ethics and drive were so humbling to watch.
I couldn’t be prouder of the Applaud team and what we achieve together. You are awesome!