Jeremy Jacobson, Developer in Atlanta, GA, United States
Jeremy is available for hire
Hire Jeremy

Jeremy Jacobson

Verified Expert  in Engineering

Data Scientist and Developer

Location
Atlanta, GA, United States
Toptal Member Since
January 20, 2022

Jeremy is a versatile data scientist with six years of experience that includes leading machine learning research projects and managing department teams. He has built data science and ML solutions used by faculty and external partners such as CareerBuilder and UPS. He was recognized by Google Education as a Google Cloud Faculty Expert and selected to the AWS Academy Cloud Council Faculty. Adept at picking up new skills quickly, Jeremy is committed to freelancing excellence.

Portfolio

Emory University (Dep. of Quantitative Theory & Methods)
Python, Bash, Jupyter, NumPy, PyTorch, SQL, Pandas, Scikit-learn, R...
Emory University (Dep. of Quantitative Theory & Methods)
PyTorch, TensorFlow, SQL, NumPy, Pandas, Scikit-learn, R, Google Cloud, Jupyter...
Toptal Client
Amazon SageMaker, Amazon Web Services (AWS), Deep Neural Networks, Data Science...

Experience

Availability

Part-time

Preferred Environment

Jupyter Notebook

The most amazing...

...thing I've worked is a GAN model that generates elliptic curves. It required new techniques for overcoming mode collapse.

Work Experience

Director of Technology

2021 - PRESENT
Emory University (Dep. of Quantitative Theory & Methods)
  • Advanced to Director of Technology after building a data science environment on AWS for two faculty and five researchers running large-scale linear job-matching models, ordered by CareerBuilder that wanted a transparent job matching model.
  • Mentored four student researchers and two faculty advisors in partnership with UPS on using the Google Cloud Platform for modeling and data engineering. The project led to UPS cutting costs and saving on labor.
  • Held three workshops and provided one-on-one training to department faculty on GPU computing. Performed the admin role on the department's GPU server.
  • Built a hosted notebook solution supporting ML workflows for our faculty.
Technologies: Python, Bash, Jupyter, NumPy, PyTorch, SQL, Pandas, Scikit-learn, R, Deep Neural Networks, Linux, Google Cloud, Machine Learning, Statistics, Data Science, Amazon SageMaker, Amazon Web Services (AWS), Algorithms, Artificial Intelligence (AI)

Lecturer

2017 - PRESENT
Emory University (Dep. of Quantitative Theory & Methods)
  • Recognized in 2020 by Google Education as one of thirty-four Google Cloud Faculty Experts. Selected in 2019 as one of eighteen AWS Academy Cloud Council Faculty, including members from MIT, Harvard, and Georgia Tech.
  • Supervised research, including an honors thesis that trains Generative Adversarial Networks (GAN) models on a novel dataset. Unlike typical GAN applications, this model generates mathematical objects.
  • Developed new techniques to overcome GAN problems such as mode collapse.
  • Designed highly lauded data science classes, QTM 250 and 350. Led over 500 students in building ML models on AWS and Google Cloud Platform (GCP) from conception through deployment.
Technologies: PyTorch, TensorFlow, SQL, NumPy, Pandas, Scikit-learn, R, Google Cloud, Jupyter, Bash, Python, Deep Neural Networks, Linux, Machine Learning, Statistics, Data Science, Amazon SageMaker, Amazon Web Services (AWS), Data Visualization, Data Modeling, Data Engineering, Algorithms, Time Series, Time Series Analysis, Financial Modeling, Amazon DynamoDB, Artificial Intelligence (AI)

ML Engineer (SageMaker) for Established AI Company

2022 - 2022
Toptal Client
  • Implemented a neural network-based time series model and trained it using custom Docker containers and AWS SageMaker training jobs. Decreased model building compute costs by a factor of 10 while enabling scalability to orders of magnitude more data.
  • Decomposed monolithic time series model code into processing, training, and deployment steps implemented in an AWS SageMaker Pipeline. Decreased lines of code by a factor of 10.
  • Mentored a team of three data scientists on using AWS SageMaker tools for MLOPs. Presented to directors in biweekly meetings.
  • Impressed the client (a large US rail company) with the project so much that they plan to use it as a building block for new work by their data science team.
Technologies: Amazon SageMaker, Amazon Web Services (AWS), Deep Neural Networks, Data Science, Data Visualization, Data Modeling, Data Engineering, Algorithms, Time Series, Time Series Analysis, Supply Chain Optimization, Financial Modeling, Supply Chain, Supply Chain Management (SCM), Artificial Intelligence (AI)

Visiting Assistant Professor

2013 - 2017
Emory University (Dep. of Math. & Computer Science)
  • Developed and managed multiple research projects, two of which delivered publications in high-ranking journals.
  • Lectured as an invited speaker at scientific conferences in England, Germany, and the USA.
  • Investigated the use of TensorFlow and deep learning on a classification problem in algebraic geometry. Ported code to TensorFlow, benchmarked models and presented this work at the Meeting on Applied Algebraic Geometry, GaTech, 2018.
  • Led a team of three postdoctoral researchers in teaching Linear Algebra and Multivariable Calculus.
  • Managed curriculum development and evaluation and was recognized for teaching excellence, progressing to Lecturer.
Technologies: TensorFlow, Deep Neural Networks, Linux, Mathematics, Jupyter, Bash, Python, NumPy, Data Science, Algorithms, Artificial Intelligence (AI)

Counting Real Roots Using Neural Networks

https://github.com/jeremyallenjacobson/RealRootsReproduction
In a .ipynb notebook file in the GitHub repository, I explain how to reproduce the results using Google Cloud, demonstrating how to use neural networks to count the number of real solutions to polynomial systems.

For details, see the slides available at http://slides.com/jeremyjacobson/deep-learning. They are from an informal talk I gave as a part of the QTM chalk talk series.

Roman Urdu NLP

https://github.com/jeremyallenjacobson/roman-urdu-nlp
A sentiment classifier that maximizes accuracy to provide a clear window into the social media conversations of the corporation's customer base. See the notebook (.ipynb file) in the included repo for details.

Model Creation and Evaluation for Sentiment Analysis

https://github.com/jeremyallenjacobson/roman-urdu-nlp/blob/master/roman-urdu-nlp.ipynb
Model creation and evaluation of machine learning models which automatically identify the sentiment of the conversations of a corporation's customer base on social media. We constructed a flexible Roman Urdu sentiment classifier suitable for multiclass or binary classification by composing existing ML models. Despite its simplicity, we have shown that its accuracy is competitive with state-of-the-art models from current research.
2006 - 2012

PhD in Mathematics and Computer Science

Louisiana State University - Baton Rouge, LA, USA

2001 - 2005

Bachelor's Degree in Mathematics and Computer Science

University of Wisconsin – Madison - Madison, WI, USA

FEBRUARY 2020 - PRESENT

Google Cloud Faculty Expert

Google Cloud Education

AUGUST 2019 - AUGUST 2022

AWS Certified Cloud Practitioner

Amazon Web Services

MARCH 2019 - PRESENT

AWS Academy Accredited Educator

Amazon Web Services

Libraries/APIs

NumPy, PyTorch, TensorFlow, Pandas, Scikit-learn

Tools

Amazon SageMaker, Jupyter

Paradigms

Data Science

Languages

Python, Bash, SQL, R

Platforms

Amazon Web Services (AWS), Linux, Jupyter Notebook

Storage

Google Cloud, Amazon DynamoDB

Other

Mathematics, Statistics, Machine Learning, Data Modeling, Algorithms, Time Series, Time Series Analysis, Artificial Intelligence (AI), Deep Neural Networks, Data Visualization, Data Engineering, Supply Chain Optimization, Financial Modeling, Supply Chain, Supply Chain Management (SCM)

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring