Tea, proper accents, and driving on the wrong side of the road are attributes which heavily represent the United Kingdom, where the team journeyed this past week to hang out with Haven OnDemand’s legendary engineers and leaders in the old and noble city of Cambridge.
Rather than take the boring way out and train it up there, I decided to live life on the edge by renting a car to drive from London Heathrow airport to my hotel in Cambridge. Luckily, I was able to get control of the reins and safely transport myself.
That weekend, we sponsored HackCambridge, University of Cambridge’s MLH event. With the UK team with us, we set off to help hackers use our APIs, and one of the oldest universities in the world did not disappoint.
Here are some of the teams that participated in our Haven OnDemand Sponsor Challenge:
Saato: Elegant ML & Data Science for Industry - 1st place winner
Saato app aim is to provide machine learning and analytics platform in the cloud without the need for data scientists. The user is provided with a dropbox folder where data can be deposited and indexed using Dropbox Cloud Connector. Furthermore the user is presented with graphical representation of trends and correlation between data sets. The user interface also provided a list of similar documents which enabled user to quickly discover highly related documents which was powered by the Find Similar API.
Named after the infamously popular meme, this team went and created internet connected billboards using a Raspberry Pi, a camera module, and our Face Detection API, helping companies advertising gain insights through analytics. The idea is simple - the more people that look at the billboard the more successful. The Raspberry Pi monitors how many people are looking at the billboard by pushing up images to our Face Detection API to verify that there was infact someone looking at the ad.
This app aims to to help people learn languages quickly and easily. Users take a photo of a an object they wish to translate. The photo goes through an image recognition to determine what it is; once determined, the name of the object is translated in the language the user wishes to learn and is run through our Query Text Index API. The user can then read some sentences about object in the language they wish to learn.
This iphone app is a smart diary. Each day, like a usual diary, you write what you did, where you went, what you ate, who you hung out with, and anything else important you did. The app then detects what language you wrote it in using our Language Detection API, finds out the overall sentiment of the entry using our Sentiment Analysis API, and extracts where you went and what celebrities you hung out with (if you’re cool like that) using our Entity Extraction API. After all is said and done, users can look back on past entries to see what what made them happiest so they can then do that in the future!
Using another team’s API which was created at the hackathon which monitors offensive words in conjunction with Telegram’s API, this team created a realtime chat application which monitors how positive or negative someone is, through our Sentiment Analysis API. Furthermore the application provides warnings whenever a person using the communicator gets overly negative or positive in order to help in steering the conversation.
The idea behind RevelApp was to create analytics platform for content creators so that they can be kept up to date with trending topics and relevant news. This app uses Cloud Web connector to index data from the website provided by the user into Haven OnDemand as well as Concept Extraction API. Concepts extracted from the website are then used as keywords for Query Text Index in order to retrieve other websites which are relevant to the one provided by the user.
The application consisted of a website where people can write an ‘About me’ section which is essentially unstructured data. This description was then indexed into Haven OnDemand and concepts extracted using Concept Extraction API and attached to the user profile. The core of the application is based on presenting a list of profiles which are refined based on concepts for the profiles which the user liked or disliked.
MarketRush is a mobile app for children to keep occupied while out shopping with their parents. The idea is to present children with an easy to complete set of riddles that can be solved by scanning the barcode of a product which the player thinks is the answer. The barcode scanning capabilities were achieved by using the Barcode Recognition API.
The idea behind CAM was to provide a map of the world with pins marking significant events which happened at particular point in the history. Users can select a particular year of interest and move through time with events being updated on the map. The event information and coordinates are harvested from respective Wikipedia pages and keywords extracted using Entity Extraction API.
A lightweight web based robot controller. It used sentiment analysis to entered text and would modify its behaviour accordingly. For example, it turned red and ran away if you said “I hate you”.
A couple of guys reverse-engineering the proprietary Bluetooth command protocol of a self-balancing robot toy they’d won at a previous hackathon. They finished this ahead of time, and so looked to extending what the robot did via the HOD APIs.
The robot had a colour-changing LED embedded in its chest, so they setup a SMS/IRC/messaging system (I didn’t quite catch what the backend was) which, when sent a message, would invoke a text-to-speech API and play this from the robot’s speaker, and also perform HOD sentiment analysis on the message and set the chest LED from bright red (sentiment -1) through white (sentiment 0) to bright green (sentiment +1).
The team freely admitted that the use of APIs was an add-on project, and their main aim (which they achieved) was to control the robot from their phones using their own derivation of the control protocol.
We saw a lot of talent and effort by the contestants. I am sure everyone learned something new and enjoyed the competition - excellent apps from all of our participants! The team that took home our Haven OnDemand grand prize was Saato.
Stay tune for a blog post from Saato.
[video demo] coding workshop
Useful Haven OnDemand Links:
- Python Client Library
- Ruby Client Library
- Node.js Client Library
- iOS Swift Client Library
- Windows 8.1 Universal Client Library
- Android Client Library
- PHP Client Library
You must be a registered user to add a comment here. If you've already registered, please log in. If you haven't registered yet, please click login and create a new account.