madewithsmiles/website

We build projects with a purpose that incorporate modern techniques and technologies.

Launchpad

We are a group of UC Berkeley students that build projects with a purpose through intelligent software.

Mission

• Solving problems with intelligent software.
• Connecting modern research at UC Berkeley with real-world applications.
• Building software focused on impact and purpose.

Curriculum

Research Areas

• Artificial Intelligence
• Machine Learning
• Computer Vision

Tools

• Amazon AWS
• Kaldi
• OpenCV

Interdisciplinary

• Student Life
• Education
• Accessibility

Spring 2017 Projects

Computer Vision: Sherlock

• Web API
  • Open-source toolkit for computer vision projects on the web.
    • Features image processing and analysis algorithms in OpenCV.

Project Sherlock is a Cloud-Based Machine Learning and Aritifical Intelligence API. We implement, optimize, and rigorously test powerful algorithms and then release our work to the public in the form of a documented API.

Speech Recognition: Luna

• Use Kaldi to train neural nets, S3 for hosting, EC2 for computation, and SQS for queuing

Speech consists of many features that are not built into personal assistants like Alexa and Google Home. We aim to build a seemless conversational personal assistant that considers prosody in speech. By feeding corpora that contain thousands of hours of annotated audio into CTC networks, our assistant will be able to adjust to speakers' uncertainty and identity.