Kintsugi, a fast-growing Berkeley-based mental health startup, is scaling and hiring for talented Machine Learning Engineer roles to help shape the future of behavioral healthcare.
Today, we want to profile one of our lead Machine Learning Engineers, Colin Vaz. Colin shares more about his journey, the problems we’re solving at Kintsugi, and the impact his work has on supporting access to mental healthcare.
Colin Vaz, PhD, a former Applied Scientist at Amazon and Sr. MLE at Behavioral Signals, combines his academic rigor, deployment experience, and adventurous spirit in exploring the new frontiers of voice biomarkers in detecting mental health conditions with the latest deep learning techniques. Colin’s research interests include robust speech recognition, speech enhancement, and machine learning.
Colin has developed ML models that have performed robustly on noisy speech and audio and in applications that include noise-robust ASR, speech enhancement, and emotion recognition from noisy acoustic and lexical information. In his free time, Colin enjoys playing piano and keyboard, and also enjoys exploring Los Angeles on his road bike.
We are excited to share this insider’s interview with you.
What inspired you to join Kintsugi?
What really drew me to Kintsugi was the chance to tackle machine learning challenges in the cutting-edge field of behavioral healthcare. With the growing challenge of mental health conditions affecting millions of individuals in the U.S., it's such an important problem to solve and the impact on society is immense. Starting with the interview process, I felt a strong sense of purpose in contributing to a technology that genuinely has the power to help countless individuals.
What project are you most proud of working on during your time at Kintsugi?
I'm most proud of working with the team to develop and optimize our stratified depression model. In classifying depression, there is naturally a scale of severity. From working directly with commercial partners, developing models beyond binary serves an important triaging function after we identify and stratify patients. Further, longitudinal models can provide additional visibility on treatment efficacy. Varied data streams from different environments from call centers to remote patient monitoring apps to telehealth platforms to clinical trials, provide additional transfer learning opportunities especially as we gather multiple clinical ground truth labels for validation. The complexity of this work satisfies my inner engineer passions, drives significant value to our business and partners, and, more importantly, has the power to transform the lives of millions who can now receive the right care in their moments of need. It’s truly groundbreaking in our experimentation cycle speed and so personally rewarding.
How has working at a mission-aligned, diverse company like Kintsugi shaped your career path?
Working at Kintsugi has been an eye-opening and exciting experience. As a machine learning engineer, it's vital to consider not only the technical aspects but also the real-world impact of our solutions. The opportunities at Kintsugi are endless. I've had the privilege to collaborate with people from various roles across the company, from Product to Clinical Ops. Everyone sees the same problem, but approaches it in different ways based on their experience. Being able to work with all of these teams and brainstorm solutions together has helped me approach problem solving more practically. It has also helped me more effectively communicate technical ideas to a non technical audience.
What words of advice would you give to new team members at Kintsugi?
Be willing to push the boundaries and don’t be afraid to explore new ideas. We are working in such a nascent category, which means there are many unknowns and countless opportunities for improvement. Fresh perspectives and ideas from newcomers are invaluable in driving our growth and innovation.
What are 3 words you would use to describe the Kintsugi culture?
Driven, collaborative, and empathetic
What is your favorite memory of working at Kintsugi?
One of my favorite memories and activities is definitely our machine learning team lunches, where we can take a break from our work to connect over lunch. These informal gatherings have been a great way to know my colleagues better in a more relaxed setting, especially when most of us work virtually.
Ready to join team Kintsugi?
Does reading this insider interview sound like you and make you excited to consider Kintsugi as your next home? Check out our open MLE roles and apply today.