(tl;dr)
I write computer code, develop hardware and design data infrastructure to better understand the world. I teach computers to do things that humans aren't so good at doing. I like finding hard problems with meaningful implications that are curious to solve.
Things I've Done Recently:
- In-house multimodal GIS data stack with harmonized ETL for the accession and processing of a heterogenous set of real world sensor and mapping data, inference and distributed content delivery
- Full-scale ownership of an embedded robotics platform for capture and analysis of high-end computer vision products for agriculture (with GIS mapping)
- Computational Neuroscience at UCSF modeling brain-wave activity in young patients with schizophrenia to predict future outcomes
- Model generation and simulation of protein interaction networks, seismic wave propagation, international supply chains
Education
Undergraduate: Stanford University, Class of 2008, B.S. Biomedical Computation
Graduate: Mount Sinai School of Medicine, 2011-2014 (3/4 years, on hiatus)
Employment
• Sound Agriculture 2023-2024
GIS Data Research Engineer III: Data architecture for geospatial products. Creating cloud solutions for routing, storing, and linking incoming data streams and creating automated knowledge discovery platforms. Layering, harmonizing, manipulation and analysis of public and proprietary GIS data sources. AWS (ECS, Fargate), IaC (Terraform), MLOps
• CloudTrucks 2021-2022
Software Engineer - Data Products: Machine learning pipelines and developing algorithms and models for data interfaces with customer facing applications. Building and scaling essential infrastructure for fleet tracking systems in the cloud. Managing resources to efficiently control data-flow for predictive models, optimization algorithms, IoT, and ML/AI. Google Cloud Platform (GCP), Django, Celery, API.
• ASCO / Cancerlinq 2021
Software Engineer: Data pipeline operations for large scale clinical data. As a consultant, I worked to extend and enhance a small-scale data environment into a robust, cloud-based architecture for continuous ingestion, processing, refinement and delivery. Java, Postgres, Airflow, BigQuery.
• AgriData 2015-2019
Data Architect / Hardware Engineer: Design, development, and management of a multi-camera, high-quality, mobile computer vision and processing robotic platform on an embedded GPU system. On the back-end, our machines were provisioned from bare-metal and my responsibilities included writing firmware, interprocess communication, database development, and linux administration. Together, they required precise TCP/IP networking, bandwidth analysis, load balancing, distributed / cloud computation, fleet management, and remote access in un-cooperative locales. In the analysis layer, I trained and maintained deep learning image models (OpenCV / Caffe / CUDA on NVIDIA Jetson TX2) for object detection and segmentation in a visually complex outdoor environment. Data input was approximately 500GB per second, per machine, and required on-the-spot processing, storage, and eventually analysis. MongoDB, Linux, C++, Azure, RabbitMQ, CUDA, PyTorch, Tensorflow, Caffe, sklearn, AWS.
Example: Leaf Segmentation [link]
Example: Fruit Detection [link]
• ProductBio / Workpology, Inc., 2015
Data Scientist: Network analysis of industrial consumer product lifecycles, quantify sustainability from seed to shelf. ETL, data ontologies, graph database design. Elasticsearch, Cassandra, Redshift, NLTK, Grafana.
• The Brain Imaging and EEG Lab at UCSF, 2008-2011
Research Associate: Using EEG and fMRI neuroimaging technologies to understand early-onset adolescent schizophrenia. Time series analysis of high frequency signals and classifications of psychiatric disorders. R, Neuroimaging, Clinical Medicine.
Research
• Case Western Reserve School Of Medicine, Dept. Of Cancer Biostatistics, 2006 & 2007
Developing models for ribonucleotide reductase, an essential DNA synthesis enzyme. To do so, we wrote software to establish parameters using data from literature, interpolating delicately when necessary, to create pharmacologic models aided by in-house algorithms based on systems of partial differential equations that replicated wet lab studies in silico.
• Virginia Bioinformatics Institute At Virginia Tech, 2005
Providing evidence, theory and software to reverse-engineer gene product interaction networks using computational power.
Personal
Born: November 16, 1985 Akron, Ohio (41.077920, -81.532937)
Current: San Francisco, California (37.780102, -122.405339)
Publications
McPherson, Selwyn-Lloyd; Zhao, Shan; Iyengar, Ravi. A Network Based Analysis of Genotypic Variation in Major Depression. Presented at Mount Sinai School of Medicine; 2012 Dec 2; New York, NY.
Colrain, Ian; Sullivan, Edith; Ford, Judith; Mathalon, Daniel; McPherson, Selwyn-Lloyd; Roach, Brian; Crowley, Kate; Pfefferbaum, Adolf. Frontally mediated inhibitory processing and white matter microstructure: age and alcoholism effects. Psychopharmacology; Feb 2011, Vol. 213 Issue 4, p669. [https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3033525/pdf/213_2010_Article_2073.pdf]
Neumann, D.A.; McPherson, S.; Klemperer, S.L.; Glen, J.M.G.; McPhee, D.K.; and Kappler, K. Documentation for a website to serve ULF-EM (Ultra-Low Frequency Electromagnetic) data to the public: How ulfem-data.stanford.edu and the servers that support it retrieve, maintain, and serve data. U.S. Geological Survey Open-File Report 2010-1321, p.49 [https://pubs.usgs.gov/of/2010/1321/of2010-1321.pdf]
McPherson, Selwyn-Lloyd. RSundials: A Suite of Nonlinear Differential Algebraic Equations Solvers in R. 2007. R package version 1.6. Available publically [https://github.com/mypolopony/Rsundials]
Skills
Computational: Python, C/C++, Java, SQL, MATLAB, R, *nix, RESTful API use/design, database management, RDF frameworks, AWS, ETL, microservices, MLOps
Technical: High throughput computer vision, Linux backends, data discovery and extraction, networks and graph databases, record linkage and identity resolution, geospatial data, schemas and ontologies, pattern recognition
Projects
Most of my previous projects are proprietary but feel free to check out Github @ mypolopony, all in various states of progress.
&tc.
Crosswords, puzzles, lemonade, and being nice to people (also, Oxford commas? Occasionally?)