Last week, Facebook held a civic hackathon, inviting in-house engineers as well as representatives from other tech companies to combine Seattle’s robust offering of open data with machine learning to solve problems.
The top teams actually came from outside Facebook. Paramita Ghosh and Raghavendra Kotikalapudi, a pair of Amazon engineers working on Instant Video, received top marks for a project that scans parking lots for available spaces, cutting down time drivers spend cruising for parking. Here is a description of the plan from Facebook.
Seattle is ranked as the fifth hardest place to find a parking spot amongst large US cities; 58 hours on average per year are wasted looking for parking. Currently, there’s no way to get good data for availability in parking lots or account for cars leaving early from pay stations. Find ‘n Park uses deep learning vision models to determine how many cars are currently parked in a given lot to give real-time availability of parking and where there might be an available spot.
Following the hackathon, the team posted an open source version of their code.
The second place team includes data scientists from HERE Technologies — a location services company owned by a consortium of Germany’s top car companies with an office in Seattle — and a Microsoft intern, among others. Their project aims to simplify the process of finding a building contractor.
When looking for a building contractor, one of the challenges is receiving a realistic estimate of the works cost. Today, you have to go through many forms before your personal information gets shared for contractors to reach out. Contractor 5 models the estimated price to complete a project by looking at both location and natural language processing to compare your project description with similar ones. This greatly simplifies how easy it is to get an estimate and increases market transparency.
Civicly-minded hackathons are far from a new concept in Seattle, but Aria Haghighi, a Facebook engineering manager who led the event said this was the first one he could recall Facebook putting on here. Almost every facet of Facebook has some representation here in Seattle, and that includes divisions focused on machine learning. There was internal interest among machine learning teams in finding ways to apply their knowledge to civic issues, and that led to the idea for the hackathon.
This isn’t Facebook’s first partnership with Seattle, as it participates in the city’s Digital Equity Initiative, as well as other programs. Haghighi said Facebook wants to be a “good citizen” in the cities it works out of, and hackathons like tomorrow’s represent another step in that goal.
