John Urbanik

Experience

otto hearing

Chief Technology Officer & Founder - Self Employed 2019 - 2020 (about 1 year)

- Developed software to improve the listening experience for people with hearing impairments. Built an at home hearing test capable of augmenting standard audiogram with information... more

HVN

Senior Software Engineer 2019 - 2020 (about 1 year)

Engineer building data pipelines and conducting spatiotemporal analysis with those pipelines. Also involved in improving search infrastructure and algorithms, performance... more

Exit

Predata

Lead Data Scientist / Data Engineer 2015 - 2018 (almost 3 years)

- Developed regularized hierarchical time-series regression + classification models using techniques from robust/causal inference for prediction of political events from web... more

Exit

Palantir Technologies

Forward Deployed Engineer 2014 - 2015 (9 months)

- Led a team that developed and E2E product that utilized gradient boosting and mixed-integer linear programming to allow non-technical users to specify constraints and preferences... more

Exit

Poptip

Software Engineer 2013 - 2014 (11 months)

- Architected real-time non-parametric Bayesian topic models for detecting changes in themes in Twitter streams.
- Implemented data pipeline running NER, spam detection,... more

Recursion

Staff Machine Learning Engineer 2021 - Present (almost 5 years)

Senior Machine Learning Engineer 2021 (5 months)

Senior Data Scientist 2020 - 2021 (8 months)

Beertending

Co-Founder / CTO 2012 - 2013 (8 months)

- Built faceted search engine and qualitative recommender system for beer.
- Developed business model and platform for crowd-sourced craft beer happy hours.

Employee

Poptip, Palantir Technologies, Predata, Beertending, Recursion, HVN, otto hearing

Education

About

Achievements

I scaled a data science and data engineering organization (Predata) 6x while developing sophisticated, explainable time-series analysis models that became integral to the product and critical to closing clients. These models operated over metadata time series (cardinality >500mm) and incorporated insights and techniques from robust statistics, causal inference, knowledge representation, spatiotemporal analysis.

Describe the most impressive thing you've done.

Skills

Python Software Architecture Distributed Systems Natural Language Processing Machine Learning Data Quality Docker Product Development SQL Business Strategy Data Analysis Performance Engineering ETL Kubernetes Computer Vision Time Series Analysis MLOps Deep Learning

Locations

New York City Palo Alto

John Urbanik

ML systems architect that employs human-centered design and focuses on applying the right tool to each problem. Deep experience in CV, time series, and NLP.