My nine-month open-source "masters" in machine learning and statistics — a full-time, self-curated schedule of textbooks, MOOCs, publishing code to my GitHub and writing to this blog — is soon ending. I'm now starting my search for an impossibly awesome "what's next."
What I bring
I bring abundant experience in: machine learning, Bayesian statistics and probabilistic programming, backend engineering, visualization and making complex ideas easy to understand.
As such, within your organization, I could immediately bring value to: ETL infrastructure, experimental design and execution, internal and external data products, i.e. algorithm design and deployment, and data evangelism efforts.
I'm looking for
- A highly technical data science role — equal parts prototyping mathematical models and deploying them to production.
- Excellent, direct technical mentorship, and the opportunity to mentor others.
- A medium-to-large-sized company with a well-established data science team. Alternatively, a smaller company with machine learning at the core of its business that, necessarily, has made significant investments in data science infrastructure and personnel.
- Large quantities of unstructured data that do not fit in RAM, ideally.
Work samples
- Minimizing the Negative Log-Likelihood, in English: Statistical underpinnings of the machine learning models we know and love. A walk through random variables, entropy, exponential family distributions, generalized linear models, maximum likelihood estimation, cross entropy, KL-divergence, maximum a posteriori estimation and going "fully Bayesian."
- Neurally Embedded Emojis: Convolutional variational autoencoders for emoji generation and Siamese text-question-emoji-answer models. Keras, bidirectional LSTMs and snarky tweets @united within.
- Random Effects Neural Networks in Edward and Keras: Coupling nimble probabilistic models with neural architectures in Edward and Keras: "what worked and what didn't," a conceptual overview of random effects, and directions for further exploration.
- Further Exploring Common Probabilistic Models: Exploring generative vs. discriminative models, and sampling and variational methods for approximate inference through the lens of Bayes' theorem.
- Dotify: A well-tested web application that recommends songs via "country arithmetic" and hand-rolled Implicit Matrix Factorization. Built with Flask, React, Webpack, PostgreSQL, Heroku and Docker.
In three years
Management, I think. I'm a strong communicator, and enjoy teaching things to anyone that will listen.
Location
NYC/SF, primarily.
About me
- I once rode a bicycle 7,614.5 kilometers from Istanbul, Turkey to Bishkek, Kyrgyzstan.
- I once taught a university Spanish course, in French, in Guinea-Conakry.
- Mathematics and code keep me smiling from ear to ear.
For more, please see my full bio and résumé. In addition, social links can be found below.