Ben Pleasanton

Dedicated data science engineer with experience in the private and public sector. Focused on creating a productive and scalable data-driven culture.

Current as of 12-20-2018

Data Engineering / Big Data

Deep Learning

Geo Spatial

Data Visualization

Statistics

Graph Theory

Data Mining

Data Governance

Business Intelligence

Software Development

Parallel Processing (HPC & Commodity)

2018 - Present
Independent Consultant - Data Science Engineer
Engineered algorithm for geospatial conflation of multiple large scale datasets (1 billion + features)
Led development on graph traversal convolutional reinforcement neural network for road detection
Architected statistical & spatial sampling library for QC evaluation of arbitrary geospatial datasets
Resolved GPU CUDA platform infrastructure whoes for internal data science research

2017 - 2018
Department of Defense
Senior Data Scientist
Developed deep learning benchmark and optimization abstraction suite including CNN, RNN, LSTM, Seq2Seq, auto-encoder, and graph algorithms to drive hardware and software scalability across multiple domains and distributed heterogeneous architectures
Developed Bayesian optimization API for scalable hyper-parameter tuning
Implemented RL algorithm for device placement using Seq2Seq based optimization of graph runtime

2015 - 2017
Groupon (OrderUp Aquisition)
Head of Data Science Engineering - Food & Drink
Engineered NoSQL-based API to create custom geographical performance visualizations on demand
Identified sales targets based on NLP analysis on reviews, cuisine categories, and restaurant types
Led highly technical team in developing model to predict delivery time customer facing
Developed kafka streaming API for near-realtime SQL based business intelligence reporting
Led team of 3 developing geo-econometric neural network in predicting geographical demand and over/under performance visualizations - Launched 2 new markets and broke volume/profit records
Pioneered ETL using Luigi data pipelines to integrate Redshift, PostgreSQL, Hive, Spark, Teradata and internal/external API data into scalable data infrastructure that remained performant and elastic
Developed Ensemble Model to predict restaurant make time and reduce costs by 20% along with increasing efficiency in the dispatcher algorithm & track performance - Patent Pending
Prototyped a simulation framework in order to test automatic dispatching strategies

2010 - 2014
U.S. Air Force
Operations Research (Officer)
Managed 12 distributed exercises per year/75-person technical team/$3M, implementing emerging technology to enhance training capabilities, and leading technical integration testing and execution for air/space/cyber weapon systems and networks to meet Combatant Commander training requirements
Developed Advanced Concepts Event tactical datalinks plan, mitigating technical network limitations to further research & development of 15 advanced weapons concepts
Performed mathematical an alysis that lead to 30% savings on $5M O&M contract
Lead CDS engineer managing $2M in DoD funding and a 16-person software development & test team with an estimated 84% cost reduction independently tested to meet performance requirements