About:
|
I like trying new things and pushing the boundaries of what we think we can do. Much of the time I find that innovation is merely the product of timely necessity. I started with a BSc. in Biochemistry & Cellular and Molecular Biology. My senior thesis included a native yeast ecology experiment with a Data Mining component using MATLAB. The abstract on the developmental basin dynamics of fermentation was published at the National Proceedings for The American Society for Enology and Viticulture in 2012.
|
|
After the first experiment I was hooked. I wanted to know more. To see the bigger picture, I needed more data. I took a winemaking internship and started to wonder what to do next. Luckily for me, that vintage on Pritchard Hill in Napa was with fruit which received a 100 point score. Perhaps more lucky than that, I was admitted into The Graduate School of Washington State University.
|
|
Long nights with the HPLC were ahead but I persevered and worked on developing better math models. I was lucky enough to meet my future wife at Washington State University. I also met a great mathematician who served on my committee for my Master's thesis using MATLAB and Machine Learning in the form of Support Vector Machines.
|
I wanted to know if we could use machine learning to predict how a wine would taste, knowing the components of the wine, and by training the machine with real human sensory data.
Each of the panelists had twenty features they rated in terms of intensity.
|
It turns out, machine learning is great at predicting stuff. The Machine Learning Thesis was published.
|
Hands On: |
I knew that in order to grow, I would need hands on experience in winemaking. Knowing the data meant joining the set. I had already been on a 100 point team. I had no idea, but I was about to be on another one. I bounced around Napa and Sonoma for four years as an Assistant Winemaker being a part of two separate, one-hundred point vintages.
|
|
|
Prototyping a blaster to create protective bio-film.
|
Recent Projects: |
Having the wine-making knowledge meant the theory was put into practice. This allowed me to examine what makes wine great. It isn't a recipe of ingredients, it's a narrative or story. Now I'm the Technical Manager for N. America at Laffort. Below, you'll find links to more of my content.
|
|
Smoke:
My most recent study is in regards to Smoke Tainted wines from N. America during the vintage of 2020. In this investigation I created a virtual wine tasting platform in February 2021 with nearly 300 winemakers as tasters. The wines were made by members of the LAFFORT technical team. The data acquisition pipeline was created using Python. Data collection was performed using smartphones and a web portal. Analysis was performed entirely using R. This study is an Exploratory Data Analysis using Principal Component Analysis, Data Interpolating Empirical Orthogonal Functions, k-Nearest Neighbors, and Multivariate Analysis of Variance using Permutations.
| ||
Enzymatics:
In 2020, I worked with the The Viticulture and Enology Research Center at Fresno State University. Here we examined the difficult to clarify Muscat and how enzymes can be customized and blended in order achieve optimal clarification, settling, or other enzymatic activity. The study used over 30 different sampling points from various vineyards in the Central Valley of California. We validated a new developmental enzyme and utilized several popular data tools for visualization with Python and ArcGIS. The study was published in peer-review with MDPI and is fully available for download under a creative commons license. The dataset is also available upon request.
|
|
Non-Saccharomyces :
Goals: This study addresses the increasing demand for ‘natural’ and certified ‘organic’ wines, along with the need for improved worker safety. Winemakers continue to search for alternatives to SO2 as an antioxidant and anti-microbial agent. This study compares the use of blended non- Saccharomyces cerevisiae yeasts, Torulaspora delbrueckii (Td) and Metschnikowia pulcherrima (Mp) as anti-microbial agents to a standard addition of SO2 on Cabernet Sauvignon. This fruit possesses over ten times the normal microbial flora typically found in California. In conjunction with this comparison study, a proof of concept prototype illustrates the use of a novel spray method for the application of these non-Saccharomyces yeasts onto a grape machine harvester for bioprotection. Published June 2020 with Catalyst: Discovery into Practice |
|
Market Penetration using Machine Learning
Clicking on the image takes you to my Github.
|
This study was based on a winery list to illustrate the ability of DBSCAN to classify winemaking regions. Coming from MATLAB, then R, and now Python has exposed me APIs capable of getting into deeper sets and diving back into Machine Learning. This is my project with DBSCAN classifying winemaking regions and illlustrates the classical traveling salesman data problem. It is a proof-of-concept study where a current client database coupled with rich API content could drive territory development and new areas of exploration for businesses seeking to optimize their time and human resources.
|
Finding All the Wineries of N. America:
Using ML to predict Wine Sensory:
|
My Master's Thesis included Support Vector Machines and Singular Value Decomposition for approximation. The main functions of this thesis were to data mine, examine, identify, and explore outliers from a sensory panel using various ethanol, tannin, and fructose concentrations in a model wine. Using higher order models provided around 93% accuracy for predicting sensory attributes in wine. Future applications of this work might include smartphone app development for consumer preference, product validation, models for optimal wine blending, robotics and sensor array development, and outlier identification for niche markets.
| ||
Sailing:
|
|
|
|
|
