Published on Tuesday, 21 June, 2022
Page is printer friendly (hopefully!)
SUMMARY
Bottom-line up front: I am an ML Engineer with a strong desire to apply state-of-the-art research to complex problems.
Recently, I've been developing a modern data platform and analytics process to identify and quantify valuable questions for
both our internal teams and our customers. My primary domain is in computer vision (segmentation/detection/classification),
but I am looking to build expertise in the domains of reinforcement learning and natural language. I am most comfortable with
python + pytorch
SKILLS
- Language: python
- Tech Stack: pytorch, conda/poetry, Azure ML, GCP, docker, git
- Development Processes: Agile, TDSP
- Certifications: Deep Learning Specialization (Coursera), ColumbiaX AI & ML, CCNA, A+
EXPERIENCE
Mile Two
SENIOR MACHINE LEARNING ENGINEER (I.E. DATA ENGINEER IN DENIAL), 2018/03 - PRESENT
- Standardized data analysis & development via a cloud-deployable conda environment to reduce spin-up time among the
Machine Learning team (targeting Azure ML) and improve data/model provenance.
- Demonstrated feasibility of metric and optimization meta-learning on customer's imagery dataset for one-shot and
few-shot scenarios
- Led a team who won 2021 and placed 3rd overall in 2019 Wright Brothers Institute's Swarm and Search AI (SSAI) challenge,
as well as received an Innovation Award for our team's vision, incorporating Deep Reinforcement Learning (DDPG)
into our submission by creating an OpenAI Gym API for the provided simulation software
- Incorporated object detection into a customer's framework. Particularly, You Only Look Twice (YOLT, a YOLO variant).
Included building streaming interfaces to feed video to the model, slicing imagery, performing non-max suppression.
Achieved roughly 11 FPS on 1080p imagery using a consumer-grade GPU. Supported distributing work to multiple local GPUs.
- Wrote technical volumes for multiple proposals focused on ML-oriented projects
PROJECTS
WBI/AFRL Swarm and Search AI Competition
2021/07
- Won 1st place overall using traditional state-based behavior typical of video game agents
2019/05
- Won 3rd place overall with a solution incorporating a mixture of classic global/local search algorithms for navigation
and deep reinforcement learning for sensor gimbal control
- Explored solutions utilizing particle swarm optimization and boid-like emergent behavior
Personal
- Open-sourced a template Dagster project to simplify development of
data pipelines. For a sample of my writing, check out the corresponding business dev blog post
and software engineer blog post
- Trained and deployed GPT2-small to generate Small Business Innovation Research (SBIR) topics and research, including
scraping and curating 27000 SBIR topics
- Completed projects for Udacity's Deep Reinforcement Learning Nanodegree, including implementing classic RL algorithms
(REINFORCE, Q-Learning) and training state-of-the-art solutions (DDPG, A2C/A3C, PPO)
EDUCATION
Wright State University
BS COMPUTER SCIENCE, 01/2010 - 12/2014
- Earned a 3.60 GPA and was present on the Dean's List Fall 2011 to Fall 2014
- ACM President from Fall 2012 to Spring 2013
EXPERIENCE (Additional)
Avatria
CONSULTANT, 05/2017 - 02/2018
- Provided ecommerce expertise to clients, including functional specification review, sprint planning, and customization
of a hybrid B2C/B2B Hybris environment
- Created a detailed action plan for developing Hybris upgrade projects, including project estimation and
requirements-gathering workshops
SapientRazorfish
SENIOR SOFTWARE ENGINEER, 01/2015 - 04/2017
- Focused on developing backend business logic and integrations for multiple short- and long-term projects for both B2C
and B2B environments within Hybris
- Led a collaborative, internal initiative utilizing popular libraries such as Spring and Redis for microservice development