You will be part of the Machine Learning (ML) team and contribute to building robust, production-ready models. You will leverage our extensive speech dataset while experimenting with a multitude of deep-learning architectures to explore state-of-the-art speech analysis methods to solve a variety of classification and regression tasks. Working alongside our cloud engineering team, you will help deploy these models and ensure they stay performant in a wide range of customer-facing applications. Minimum Qualifications • M.S./Ph.D. in Computer Science or equivalent or B.S. with 5 years of experience in building production-grade machine learning models in industry and/or academic research settings • Strong programming skills in python with extensive experience with the scientific and deep-learning stack (numpy, pandas, numba, torch, tensorflow, jupyter) • Background in speech processing or audio classification • A proven track record of building end-to-end neural network models and presenting results to colleagues • Experience optimizing the compute performance of models for production • Ambitious team player with strong communication skills (oral and written) • Experience implementing and experimenting with cutting-edge ML techniques from the literature Kintsugi is on a mission to scale access to mental healthcare for all. We are developing novel voice biomarker software to detect signs of depression and anxiety from short clips of free form speech. Awarded multiple distinctions for AI technology and recently named one of Forbes' Top 50 AI companies to watch in 2022, Kintsugi helps to close mental health care gaps across risk-bearing health systems, ultimately saving time and lives.

Mental health issues have come into a clearer focus amid the pandemic. Depression became endemic, but it still too often goes undetected. Even when it does, healthcare providers struggle to meet demand. Two women engineers -- both of whom experienced depression and had trouble finding therapy -- thought the answer might be helping medical pros detect depression. Kintsugi is a startup that wants to put technology to work on the problem.

This project is an exploration of different stacking methods of fragments using computation, and comparing the Hungarian algorithm with Minkowski difference Algorithm in nesting items. A python code on Grasshopper was scripted to provide a method of iteratively stacking fragments one-by-one. Using the Minkowski Difference Algorithm (a simple subtraction of vertices), a path to move polygons around without collision is generated, allowing us to stack fragments together without overlap. This method is already used in graphic modeling for detecting collisions and nesting applications such as SVGNest by Jack Qiao. Many parameters such as rotation, trying different orientations, were explored to find a more efficient method of stacking the fragments.

Making mental healthcare easily accessible to anyone is what led Rima Seiilova-Olson (MSTM '10) to become co-founder of Kintsugi Mindful Wellness, talk therapy software that combines machine learning and voice journaling to tackle stress, anxiety, depression and loss. "There's a big opportunity right now to use artificial intelligence for good. AI is not'summoning the demon' like Elon Musk says. When you're suffering, you need affordable access to help right away," said Seiilova-Olson, who met Kintsugi cofounder Grace Chang at an OpenAI hackathon in San Francisco. They quickly discovered they shared a passion for exploring how technology can help people address their mental health needs. "Some people have expressed their doubts about our idea, but I learned at Gies Business that good ideas often are met with a lot of resistance," said Seiilova-Olson.