Hey, I'm Jon 👋

I am a Neuroscientist 🧠 turned Data Scientist 👨‍💻 with 10+ years experience using machine learning to make data-driven discoveries and deliver business solutions.

headshot
doorway

What I do

Full stack data scientist with expertise in client-facing interactions and MLOps tooling.

  • python

  • GCP

  • docker

  • SQL

  • aws

⚡ Process-driven data scientist.

⚡ Well-versed in client-facing interactions.

⚡ Experienced at building scalable production-level deployments.

Experiences

Databricks
Databricks
Specialist Solutions Architect
July 2024 – Present

Machine learning + GenAI major and data engineering minor. Notable projects include:

  • Scaling forecasting training and inference workloads for optimal performance on 100,000+ models
  • Distributed training and inference of large-scale forecasting across multi-node clusters
Beyond Limits
Beyond Limits
Senior Data Scientist
Nov 2021 – July 2024

Data science lead for scoping POCs and bringing MVPs to life. Projects include:

  • Battery life-cycle prediction to reduce testing time
  • Sand management advisor for British Petroleum
  • Lead instructor for problem discovery and solution building for Aramco x Caltech AI Academy
SoCalGas
SoCalGas
Data Scientist
Aug 2020 – Oct 2021

Modernized antiquated workflows and founding member of SoCalGas' Model Review Board. Projects include:

  • Daily gas load forecaster application for internal gas acquisition
  • Route optimizer for gas meter inspection
University of Southern California
University of Southern California
Doctoral Researcher
Sep 2014 – Apr 2020

Completed dissertation on "The cortical representation of touch". Experience include:

  • Open-sourced transfer-learning application for touch-frame detection
  • Data pipelines to clean and synchronize 30 million timepoints of sensor motion and neural recordings
  • Feature engineering pipeline using time-series filtering and physics models

Education

University of Southern California
University of Southern California
Doctor of Philosophy; Neuroscience

Sep 2014 - Apr 2020

Completed dissertation on "The cortical representation of touch"; published three peer-reviewed articles:

  • Behavioral and neural bases of tactile shape discrimination learning in head-fixed mice. (Neuron 2020)
  • Active touch remaps barrel cortex output from a representation of self-motion to object location. (PLoS Biology 2020)
  • The behavioral basis of whisker-guided anteroposterior object localization in head-fixed mice. (Current Biology 2019)
University of California, San Diego
University of California, San Diego
Bachelor of Science; Human Biology

Sep 2009 - Apr 2013

Foray into neuroscience research and student body representative

  • College council president (2011-2012)
  • Biological sciences senator (2012-2013)
  • Leutgeb lab studying learning and memory (2011-2013)

Open Source Projects

databricks-bookshelf

🧱 An exploration of the Databricks ever-growing feature set.

Python

0

1

1 MB

ML-deployment

Local machine learning deployment leveraging MLflow, Optuna, and FastAPI.

Jupyter Notebook

1

1

1 MB

scaled-ML-deployment

Kubernetes scaled machine learning deployment.

Jupyter Notebook

0

1

328 KB

Pub_LocalizationBehavior

Code associated with the figures of Cheung et al.

MATLAB

1

1

295 KB

Pub_S1LocationCode

Data and code associated with the Cheung et al., 2020

MATLAB

0

0

42.1 MB

Blogs

Reflecting on my doctorate

I believe this is the key component to solving innovative problems quickly and growing. To those in software development, methods for organization are defined as agile methodologies and frameworks such as Scrum and Kanban. These frameworks highlight clear and actionable goals with daily, weekly, and monthly milestones. With clear milestones operating on short timescales adopters of this principle, whether they be individuals or in teams, adapt and grow quickly.. As a doctoral student and those in industry that are innovating, the problems you are solving has probably never been tackled before and the steps to reach your goal are likely nebulous. With clear project goals but nebulous steps, frameworks for planning and flexibly improving become ever so important. For myself, I spend the first Sunday of each month planning out goals for each week and every Sunday planning out goals for the daily. Planning involves writing down goals sorted by order of importance with daily, weekly, and monthly deadlines. Set goals that are both practical to reach and also a bit further than you could’ve imagined yourself reaching for. The process of writing down goals mentally prepares your mind so that going to work doesn’t come as a shock. Organize and deliver and you’ll go from treading water to swimming with the current.. My doctorate taught me that the best teams aren’t full of similar individuals. Yes we all know this saying but where that difference should lie was something I had to learn. Hire too diverse a set of individuals and collaboration will be difficult. Hire too similar of individuals and there’ll be no creativity.. As the first hire, I had a say in choosing who would join our team. I selected individuals who had the same disciplines in organization, as I believed this was the defining trait that made a doctorate successful. However, my manager vetoed many of my decision to pass on individuals and I thought he was dead wrong. After years of working with my team I saw how we devised creative solutions, inspired growth amongst one another, and made fascinating discoveries. We would each spend hours working on our problems and at the end of the day or the week we’d stay after hours, crack open a can of beer, and talk science over great food. These conversations forced each person to clearly communicate their ideas in front of peers that were highly inquisitive and had different ways of thinking. Ideas were refined after each shared meal and we all grew to love having our ideas challenged.. Looking back I saw that I was thinking too small in what defined a great scientist and what to look for when drafting a successful team. What I look for may broaden in the future but I’ve learned to embrace individuals who are inquisitive and will chase problems down. I believe a team of individuals who embody this simple truth will bring success to any company and mission.. As years passed and many conversations were shared over meals and hallway passings, I saw how my thought process was being refined.. The biggest challenge I had to overcome was learning humility and receive criticism well. Solving complex problems required not just grit but the forging fire of individuals who were unafraid to question your every step. I was humbled by the process of angrily defending what I felt were fully fleshed out ideas but still having my team continually pour great ideas onto me. The click happened when I recognized their questioning came from a place of curiosity and not malice or pride. Standing my tallest and seeing loving teammates pour great ideas onto me was a truly humbling experience.. As an example, when it came to building a predictive behavioral model for my first major research project the difficulty came in creative feature engineering and model interpretation. It wasn’t good enough to build a model that predicted well but one that could be understood and press that boundary of what we knew about the brain and behavior. Defining which features were most important and which machine learning tool was best fit required pooling knowledge, experience, and intuition from the team. Eight months of frustrating conversations resulted in a refined model where we, as a team, knew the ins and outs of each intricate component and our first major publication in Cell Press.. As taxing as this process was, I wouldn’t trade those moments and conversations for anything. Today, iterating ideas and making discoveries with the team is a much quicker process. Laying my pride down (though it’s still hard to do) lets me receive feedback and grow ideas. Without them for me and me for them, our team would not have been as productive as it is. I love the people I work with and my thought process as a scientist would not be what it is without them.. I’ve grown to love the place I’m in but I recognize that there is still plenty of room to grow and learn. This desire to constantly grow could be a by-product of what a doctorate does, where over the past half decade I’ve learned to question everything including myself. I always find room for improvement and I always look for ways to learn new things. Whether it’s organizing and setting goals or conversing with others about my ideas, everything has been pointed towards sharpening me and challenging the way I think. Regardless of where I go next, I look forward to still learning..

Reach Out to me!

Discuss a project or just want to say hi? My Inbox is open.

"Specialist Solutions Architect @ Databricks | Ph.D. Neuroscience"

Open for opportunities: No
Jonathan Cheung