“Scott is a true ninja, both metaphorically and physically (you should see his moves). Having known Scott in college and now having had the chance to work along side him, I cannot recommend him enough. He defines two very important words in a colleague: dedicated and loyal. Both of these bleed into everything he does. I'm not sure how, but day in and day out, Scott found a way to operate at anywhere from 106-120% and do more than any human should be able to. I loved working with Scott and appreciate everything he brought to the table in his personality and work ethic. I know our paths will across again in the future and can't wait for it!”
About
The first time I crossed the Golden Gate Bridge, I had 63 days and 2,000 miles of cycling…
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
Miami University
-
-
Activities and Societies: Pi Kappa Phi, IFC Community Service Board, Greek Life Guide
Licenses & Certifications
-
-
Data Science Specialization
Coursera Course Certificates
-
Developing Data Products
Coursera Course Certificates
-
The Data Scientist’s Toolbox
Coursera Course Certificates
-
-
-
-
Exploratory Data Analysis
Coursera Course Certificates
-
Getting and Cleaning Data
Coursera Course Certificates
-
Volunteer Experience
-
Journey of Hope Cyclist
The Ability Experience
- 3 months
Cycled 4,000 miles from San Francisco, CA to Washington D.C.
Enjoyed incredible friendship visits daily at camps and organizations for people with disabilities.
Surpassed personal fundraising goal by 20%, raising $6,000.
Participated in programs to raise awareness for people with disabilities.
-
Projects
-
Enron Persons of Interest
See projectFor this project in the Udacity Data Analyst Nanodegree I explored if I could use machine learning to find which employees at Enron were "persons of interest" for the fraud that took place there.
While I have done about twenty projects for various online courses, this is the project where I learned the most by a large margin! From picking up new tricks for data exploration in pandas, learning to use sklearn, and greatly improving my understanding of tuning and validating machine…For this project in the Udacity Data Analyst Nanodegree I explored if I could use machine learning to find which employees at Enron were "persons of interest" for the fraud that took place there.
While I have done about twenty projects for various online courses, this is the project where I learned the most by a large margin! From picking up new tricks for data exploration in pandas, learning to use sklearn, and greatly improving my understanding of tuning and validating machine learning algorithms, I was thrilled to practice such a wide variety of skills. -
RxNormR
See projectWhile volunteering at SF Brigade we developed an R package for accessing the National Institute of Health’s RxNorm database.
I was responsible for writing most of the functions for calling the API and their documentation.
This made it much more convenient to access the database for another project we were exploring.
Honors & Awards
-
Kaggle.com Script of the Week
Kaggle
-
Eagle Scout
Boy Scouts of America
Recommendations received
1 person has recommended Scott
Join now to viewExplore more posts
-
Elliot Marx
Since Databricks announced its acquisition of Tecton last summer, I’ve been hearing from a lot of ML teams who feel uneasy. They’ve spent years investing in Tecton’s feature platform, now they’re staring down tighter lock-in and uncertainty about what happens next. The good news: there’s a better path forward. Chalk runs in your cloud, connects to your existing data ecosystem (Snowflake, BigQuery, Postgres, Kafka… wherever your data lives), and keeps your compute and serving choices open. No lock-in, no forced migration. On Thursday, Melanie and I are hosting “Life after Tecton: Migrating to Chalk,” a live session on how top ML teams are preparing for what’s next and taking back control of their stack. If you care about interoperability, speed, and ownership, this one’s worth joining. Register in comments!
3 Comments -
Rose Tyler Becerra
I enjoy using Claude to normalize and analyze large sets of public data. Often this is after I've wrestled with the raw data in a spreadsheet enough to know what patterns are there and where I think bodies are buried in poor data quality. It's validating to go from the feeling "this data is crap" to the fact that "16% of the data has a key field missing." I could do anything with AI but here I am amplifying my data nerd powers.
-
Leo Delmouly
Heard this quite often in the past few months: get around Snowflake lag by... ...pushing analysts to #streamprocessing abstractions. From what I've heard, this is a recipe for stalled adoption. When Snowflake's batch refresh lag feels too high for an analytical case that's meant to be "real time", the solution becomes moving analytics closer to the stream. #streamingSQL enters the picture (FlinkSQL, ksqlDB, etc…) because it looks familiar enough to feel safe, but it's misleading. Snowflake trains people to think in snapshots. You ask a question, you get an answer that represents a moment in time, and you trust it. If the answer looks odd, you change the query and run it again. Nothing keeps running after you close the tab. Stream processing doesn't work that way because time stops being implicit. You're no longer running queries - you're deploying jobs that continuously process data. Instead of just asking "what is true now", you're deciding how that truth should change as late data arrives, as events come out of order, or when upstream sources send corrections. That shift is easy to underestimate, and a direct consequence is that iteration slows down. In Snowflake, a bad query just returns a bad result. In FlinkSQL, a bad query can stop a job that's been running for weeks - and someone has to decide whether to restart it, backfill missing data, or accept that the continuity broke. And it's not just that mistakes are more expensive - they're easier to make. Analysts who write correct batch SQL queries can easily write streaming queries that look right but produce subtly wrong results because time semantics are implicit in ways they're not used to thinking about. People stop making changes because the blast radius got bigger. What happens next is that a few people end up owning the streaming stuff, everyone else goes through them, and when someone needs a fast answer they just use Snowflake anyway. The answer probably isn't more FlinkSQL education. It's giving analysts an interface that preserves how they already think - tables they can query without caring about stream semantics. Not sure stream processing concepts are the right interface for analysts in the long run. Even with managed services handling the operational complexity, when the goal is fresher data, moving bytes faster is the easy part. Not changing how people work is the hard part.
1 Comment
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content