We’re excited to kick off the 1st Sourcegraph Open Source Fellowship, a program where we sponsor talented college students to work on (or start) any open source project. At Sourcegraph, we are committed to helping more people benefit from the open source community, so we’re extremely happy to give students the opportunity and the mentorship to help them pursue their interest in open source. We received tons of great applications and selected 10 local SF Bay Area students with a variety of backgrounds, from first-year CS majors to masters students.
Join us in welcoming the Sourcegraph Open Source Fellows! All 10 fellows’ bios and project descriptions are below. The program’s just starting, so not all projects have links yet. We’ll add links and feature each project in more depth over the next 2 months on this blog, so check back and follow @srcgraph on Twitter.
Interested in participating in the next Sourcegraph Open Source Fellowship, tentatively scheduled to begin in April 2014? Sign up to get notified when details are released.
Andy Li is building a project (codenamed Miditype) that allows for computer keyboard input to create MIDI sequences for musical composition, education, and enjoyment. Whereas existing music notation software requires precise inputs and special hardware, such as a MIDI keyboard, this project will allow users to use a standard computer keyboard to input notes. Users can play music just like an instrument and the program will make automatic improvements on the input. The goal is to make it easier to compose music on a computer.
Andy is an EECS freshman at UC Berkeley who has also taken independent machine learning courses on Coursera. He’s excited about applying his programming skills to the world of music and about contributing to the open source community.
Ansh is a Stanford Math/CS sophomore whose favorite classes so far have been compilers, networking, and algorithms. On campus, he is an active member of BASES and is on staff for the Stanford Flipside. He’s interested in the challenge of building a real, useful implementation of a complex algorithm. Layouts and visualizations of DAGs are important for organizing human knowledge, which is one of Ansh’s deep interests and long-term goals.
Gabriel Bianconi is developing django-google-prediction, a Django application that wraps the Google Prediction API to make it easier to build sites with machine learning or data analysis functionality. The project will simplify the integration of the Google Prediction API with Django’s models and views.
Gabriel is a Stanford freshman who has taken courses in Java, scientific Python, and C++, in addition to Udacity CS courses. He also has independent web development experience, using Python, Django, HTML, CSS, etc.
Jay Hack is developing NIPy (Natural Interaction for Python), a library that makes it easy for users to gather data from devices such as the PrimeSense and Leap Motion for use in inference tasks, such as gesture detection and activity recognition. Few of these devices provide support for Python or offer high-level abstractions for inference tasks, which Jay’s library will provide. This project was inspired by his experience on 3 prior natural interaction projects, including a 2nd place finisher at the Yahoo Summer Intern Hackathon.
Jay is a Stanford CS junior and has conducted independent research on machine learning and computer vision in the AI lab of Prof. Ben Kuipers (University of Michigan). He’s interested in continuing to apply machine learning to natural interaction (NI) data and contribute to the NI community’s knowledge base and open source toolkits.
Julia is helping Stanford public health researchers in Bangladesh quantify exposure factors for childhood diseases to better identify causes of health risks such as diarrhoeal disease (the second most common cause of death for children under 5). She’s partnering with these researchers to develop an app for use on their Android tablets to enable easy mobile video coding (hand-labeling of scenes). This app will help the researchers measure exposure factors such as frequency and duration with which a child’s hand and/or mouth contacts soil, feces, or animals, and it can also be applied to other video coding tasks, such as wildlife observation, psychology studies, and healthcare training.
Julia is a Stanford CS masters student and has interned at Google, Palantir, and Khan Academy. She’s excited to contribute to open source projects for social good!
Kelsey Josund is building a web application to help people find information about U.S. congressional incumbents. It will show the incumbent’s congressional district, the date of their next election, and the margin by which they won their most recent election. The app will help constituents and campaign members determine how firmly an elected official is in possession of their seat. Low victory margins or an upcoming election indicate that someone is more likely to be swayed by their constituents.
Kelsey is a Stanford CS sophomore and has taken CS classes involving Java, C++, theory, probability, etc. She’s interested in building API-driven applications and working on independent, large-scale projects in addition to her Stanford coursework.
Kevin Lin is working on java-audio-steganography, a project he made that encodes data in analog audio. He’s improving the codebase, adding support for non-textual data, creating a test suite, and reaching out to folks who have expressed an interest in the project to solicit community contributions.
Kevin is an EECS freshman at UC Berkeley. He learned Java when he was a freshman in high school and has taken the intro CS class at Berkeley.
Leo is a EECS freshman at UC Berkeley who has been working as a freelance web developer/designer for 4 years and hacking full-stack at startups for the last 2. He has experience with Node.js and has also worked with PHP, Rails, and various SQL/NoSQL databases.
Minyoon is building a patent search tool, PatHunt, to help attorneys and inventors discover prior art based on explicit references and implicit topic relationships (using, e.g., term co-occurrences). In addition to parsing patent metadata and texts, the project helps people visualize the web of related documents for a given patent. Minyoon is getting help from Prof. Lee Fleming of UC Berkeley, who studies innovation and entrepreneurship.
Minyoon is a CS senior at UC Berkeley with an interest in patents and data science. He’s interested in building an open source project and participating in the open source community.
Stephen Macke is building Jaydio, a Java library for Unix variants to perform low-level file I/O which goes direct to disk, bypassing the file system cache. Existing Java direct, random-access I/O routines are lacking. If time permits, he’ll use his library to add optional functionality to the low-level file I/O routines of the Apache Lucene project. This project will use JNA to wrap libc functions to perform the I/O operations.
Stephen is a first-year Stanford CS masters student focusing on AI and systems. He is conducting independent research with Prof. Jure Leskovic (on graph algorithms for the Snapworld project) and was an ICPC world finalist in 2012. Stephen interned at Palantir last summer, where he worked with Lucene and other Java technologies.