Charles Tripp Email and Phone Number

At NREL, I research machine learning techniques and apply them to renewable energy and energy efficiency domains. I work in PI, co-PI, key contributor, and advisory roles on across multiple projects. I am a dedicated advisor to early-career scientists, postdocs, and grad students. My work bridges the gap between emerging AI research and real-world applications. I specialize in empirical machine learning, reinforcement learning, probabilistic modeling, and research at scale utilizing supercomputing systems. I have utilized over 250 node-years of computing time across five different high-performance computing systems. Applications include connected autonomous vehicle control, hybrid solar-wind plant layout optimization, short-term wind prediction, augmenting traditional control techniques with RL methods, and behavioral modeling of golden eagle flight patterns. I have lead and participated in more than fifty proposals, with funds secured in excess of $8.3M.

Creating the integrated algorithm management system of the future. I founded this venture-backed enterprise software startup to create a new category of software: the algorithm management system. We are continuously growing our product, revenue stream, and client list. Our product enables marketers, CEO's and CMO's, managers, data scientists, and engineers to create, tune, explore, deploy, and manage the most critical algorithms that their businesses depend upon. Creating a system which makes query building and management fun, easy, and approachable for non-technical users is a challenging and engaging problem. As CTO, I have had the responsibilities of executive, director of research and development, director of engineering, lead researcher, and project manager.Terrain was successfully acquired by Confluent, Inc in 2018.

I led research and development of technologies that help users find products they wish to purchase. Application areas include image search, text search, and recommendation systems. I created and patented an image matching search engine, combining new and existing image processing techniques to compare both the whole of images and their component parts. Both near-identical and semantically similar images can be discovered with a high degree of accuracy. I developed the image processing algorithms, as well as the information retrieval system itself. My techniques were patented in 2013.

As the first employee, I helped raise $5 million in funding for this recipe search engine startup. I led the research and development of a novel natural language search engine for finding recipes. This high-performance search engine uses Bayesian techniques to sort and filter recipes in response to a text query and recommended recipe queries. The system additionally provides automatic hierarchical clustering and navigation of result sets. The Natural Language Processing system powering text analysis uses a low latency linear time parsing algorithm which is able to infer large amounts of information from recipes and text queries which is then used to better search, filter, and categorize the results. Our search engine was successfully sold to InterActive Corp. in 2015.

Management Science & Engineering, Decision Analysis 152 and 252. Working as a team with other course assistants, I helped students become familiar with the basics of decision making, and eventually to become expert decision makers.

I researched, designed, and implemented a high availability, scalable information retrieval platform for recommendation and search using kernel-space comparisons and locality sensitive hashing. I designed and implemented a data synthesis engine which gathered user predictions and demographic information to produce insights into the thoughts of the predictors.

I collaborated with Andrew Ng and Pieter Abbeel to increase fuel efficiency during formation flight using reinforcement learning and Kalman filtering. Both Andrew and Pieter are leading researchers in reinforcement learning and deep learning techniques.

I worked as the lead designer and programmer for a complex, fully-programmable, database performance simulator. I built the event-driven architecture and a LISP-style interpreter from the ground up in a fully object-oriented C++ design. I used ANSI C++ for use in both Windows and Linux. Q-Sim continues to be developed for both internal and customer use in capacity planning and design tool roles.

I worked with test, verification, and design engineers to perform the majority of the switch-level functional and physical verification of a complex multiprocessor controller chip. I created software to automate and parallelize the chip verification process.

I worked with the Colorado RF Engineering department, performing many of the duties of an RF Engineer. I assisted engineers in predicting future network traffic, planning the placement of new cellsites to expand and enhance the network, troubleshooting sites, maintaining cellsite databases, and collecting drive test data.