Gunnar Pope, PhD

SELECT EXPERIENCE

Energy Market Forecasting (DA/RT LMP, Net Load, Solar, Wind) using AI/ML, Markov Chains, and Probabilistic programming in Python; Dispatch Optimization using Reinforcement Learning, Bayesian Optimization, and Dynamic/Linear Programming; Digital Signal Processing; Cloud-Computing on AWS, Embedded Design in C, MLops, IoT, Test-Driven Design, Automation

CURRENTLY

Founder: BitStory.AI 2022-Present
Leveraging data to decarbonize the planet. Energy market forecasting and dispatch software to minimize financial risk and carbon emissions for renewable energy systems. Specializing in probabilistic, time-series forecasting software to help businesses make better decisions under uncertainty.

PROGRAMMING EXPERIENCE

Python 2017-Present
Machine Learning (Scikit-Learn, PyTorch, Keras, Tensorflow)
Probabilistic Programming (pymc, Tensorflow Probability)
Time-Series Forecasting (ARIMA, LSTM, GRU, CNN, RNN)
Bayesian Analysis, Inference, and Optimization (PyMC3)
Data Visualization (Matplotlib, Seaborn, Plotly)
SQL and NoSQL Databases via AWS Boto3 (PostgreSQL, DynamoDB)
Test-Driven Design (unittest)
Data Collection (ETL) and Automation via AWS Lambda
Package Management (conda, pip)
Wrapping & Testing Applications Written in C
Object-Oriented Programming
CI/CD Pipelines

AWS Cloud Computing 2020-Present
Serverless: AWS Lambda
Databases: AWS S3, DynamoDB, RDS Websites: AWS ElasticBeanstalk

C 2012-Present
Ultra-Low Power/Low-Latency Applications
Embedded Machine Learning Advanced Digital Signal Processing
Data Compression: Lossless and Lossy Filters: Kalman Filters, IIR, FIR Filters Transforms: Fast Fourier Transforms, Wavelet Transforms
Wireless Communication (BLE, Zigbee)
Injected Dependencies
Test-Driven Design via Unity Test Suite
Event-Driven Design

Linux 2012-Present
Ubuntu Desktop
Ubuntu Server
Debian (RaspberryPi)

Bash 2012-Present
Scripting and Automation

Docker 2018-Present
Containerizing Applications
Infrastructure-as-Code
Modular Application Design
CI/CD

EDUCATION

Dartmouth College, Hanover NH 2014-2019
Doctor of Philosophy (Ph.D.)
Thayer School of Engineering, Hanover, NH
Thesis Title: An Ultra-Low Resource System for Electrodermal Activity Monitoring
Thesis Topic: Ultra-Low Power Wearable Biosensors
Concentration in Electrical Engineering
Advisor: Ryan J. Halter, Ph.D.
Relevant Coursework: Ultra-Low Power Embedded Systems; Advanced Digital Electronics and Signal Processing; Biomedical Instrumentation Design; Computer Engineering; Machine Learning

Dartmouth College, Hanover NH 2012-2014
Bachelor of Engineering
Thayer School of Engineering, Hanover, NH
Major: Electrical/Electronics Engineering
Citation for Meritorious Performance, Dartmouth College Nov 2013
Relevant Coursework: Microprocessors in Engineering Systems; Power Electronics; Analog and Digital Design; Distributed Systems

Flathead Valley Community College, Kalispell MT 2010-2012
Associate of Science in Engineering
Concentration in Electrical Engineering

University of Colorado, Boulder CO 2001-2005
Bachelor of Arts Major: Physical Geography
Minor: Atmospheric Science

PROFESSIONAL EXPERIENCE

BitStory.AI 2022-Present Hanover, NH
Founder/Algorithm Engineer of BitStory.AI-Leveraging data to decarbonize the planet. Energy market forecasting and dispatch software to minimize financial risk and carbon emissions for renewable energy systems. Specializing in probabilistic, time-series forecasting software to help businesses make better decisions under uncertainty with deep expertise in modeling discrete and probabilistic systems. Currently developing microservice architectures based upon AWS Lambda, S3, and EC2 technologies for continuous forecasting and dispatch optimization. Expertise in high-bandwidth / low-resource applications for applications in energy, cyber-security, and edge computing.

Simbex LLC 1/2020-9/2022 Lebanon, NH
Lead Engineer designing wearable biosensors for physiological monitoring. Working with clients to translate their product vision into technical requirements and features. Leading hypothesis-driven testing for product performance evaluation and quality control. Digital architect for real-time processing applications on resource-constrained devices. Leveraging machine learning (ML) algorithms to classify physiological behavior. Deploying Machine Learning applications (MLops) into the cloud and on embedded devices. Automating development pipelines (unit tests, CICD, etc.) for high-quality, continuous deployment. Full stack algorithm development using Python for algorithm development. Translating algorithms from Python into embedded-C on a wearable, wireless device. Extensive background in low-power, wireless communication (BLE).

Ph.D. Candidate and Researcher 2014-2019
Thayer School of Engineering at Dartmouth College, Hanover, NH
Collaborating with Dartmouth’s Computer Science Department and the Center of Technology and Behavior Health translate digital health technologies into real-world applications. Developing analog and quasi-digital instrumentation for ultra-low power biomedical sensors. Specializing in digital signal processing on low-resource embedded devices. Applying full-stack computer engineering methods to measure, record, visualize, and analyze biomedical data. Characterizing of sensor performance using statistical analysis.

Quasi-Digital Sensor Design 6/2019
First development of a wearable, long-term physiological recorder based on a single 16-bit microcontroller. Implemented a multi-level wavelet transformation of the EDA signal to compress the physiological data by 23x. Applied real-time signal compression algorithm on-board a 16-bit MCU.

Cook Engineering Design Fellow 6/2014–9/2014
Thayer School of Engineering at Dartmouth College, Hanover, NH
Recruited engineering projects from industry professionals by matching workplace problems with relevant student talent. Defined scope and breath engineering projects for ENGS 89/90 program. https://engineering.dartmouth.edu/cook/

Energy Analyst Intern 6/2013-9/2013
Energy Program Office at Dartmouth College, Hanover, NH
Created tools to automate analytics for industrial-scale HVAC data sets to pinpoint inefficient processes. Constructed and tested efficient collection regimes for large-scale data sets. Communicated results to team leader and presented recommendations at weekly meetings with colleagues.

Owner 2006-2012
Pope Custom Construction
General Contracting, Whitefish, MT
Managed, supervised and constructed numerous multi-million dollar homes. Coordinated tradesman across multiple stages of the building process.

RESEARCH INTERESTS

Forecasting Distributed Energy Resources (DERs), Probabilistic Forecasting using Neural Networks, Applied Machine Learning Applications for Health and Wellness Technologies, Stress Detection, Gait Analysis, Low-Power Sensor Design, Instrumentation Design, Algorithmic Trading

PROFESSIONAL ACHIEVEMENTS

Patent Pending Aug 2018
``A micro-recording device for physiological signals”
Inventors: Halter, R.J., Pope, G.C.
Application Number: US17/270,971
Publication number: US20210251574A1

Research Funding Oct 2016 National Science Foundation, Computer and Network Systems
Award Number: CNS-1619970- Award Amount: $815,840.00-PI(s): Halter/Kotz

Research Funding Oct 2014
National Science Foundation, Computer and Network Systems
Award Number: CNS-1314281 & Award Amount: $1,009,430.00
Award Number: CNS-1314281-Award Amount: $1,009,430.00-PI(s): Halter/Kotz

Fundamentals of Engineering Exam May 2014 National Council of Examiners for Engineering and Surveying, Concord, NH Certification Link

PROJECTS

Net Load Forecasting Competition Jul 2023

• Day-ahead forecasting competition of net electrical load at 4 different locations for 30 days with high renewable (solar) generation.
• Final ranking was Top 15 overall (competing against commercial forecasters and universities with 4+ team members)
• Utilized a variety of ML/AI technologies for probabilistic forecasting models based upon RNNs, CNNs, and Gaussian Processes
• Automated daily submission of 24-hour probabilistic, net-load forecasts to remote client
• Leveraged serverless API framework to gather global, national, and regional meteorological and climatological data
• Developed automated ETL pipelines for weather and energy data to forecast net load

Web Sensing: High-Speed Grammar Parser 2022-2023

• Designed a “Sequence Combinator” (i.e., lossless compression algorithm for sequence-based data) for a Push-Down Automata used to parse an LALR grammar for a high-speed, internet security applications.
• Combinator Parser Results: 90% improvement in BRAM utilization, and +10% Mbps improvement in bandwidth over traditional parsers Bison Parsers, and overall 95% reduction in parser table size.
• The Sequence Combinator results enable the real-time parsing of JSON objects over HTTP between a server and a client, at internet speeds, such that malicious data packets are discovered and dropped before being transmitted to the client.

Silvertree: Fall Detection Sensor 2021-2022

• Lead Algorithm Engineer designing an embedded, fall-detection algorithm for aging adults
• Algorithm embedded AI/ML (Decision Trees) within an embedded IMU (LSM6DSOX) for ultra-low power performance to detect potential fall events. Raw data was post-processed on the MCU for final inference of fall/no-fall events.
• Designed the data-driven, experimental methods for algorithm development and verification
• Leveraged wireless BLE communication using Nordic Semiconductor’s BLE SoC (nRF52840)
• Embedded ML algorithm withing STmicro’s LSM6DSOX for ultra low-power operation
• Link: Silvertree

Medrhythms: Embedded Gait Analyzer 2021-2022

• Lead Algorithm Engineer: Designed a real-time gait cycle time analyzer
• Algorithm designed leveraged 3-axis accelerometry and 3 axis-gyroscopic data to infer the step frequency and stride length of recovering stroke victims, in real-time.
• Gait detection algorithm measures time of Toe-Off (TO) and Heel-Strike (HS) events from raw IMU data with <30ms error-competitive with Physiolog’s state-of-the-art Gait Up sensor.
• Ported gait analysis algorithm from Python into embedded C on a wireless BLE (nRF52840) chip
• Link: Medrhythms

OPOS1 Compliance Monitor 2020-2021

• Designed a complete IoT product within 6 months to monitor usage of orthotic/prothetic limbs
• Designed and produced a sticker-like wearable sensor that attaches to a prothetic limb and a clinical application
• Ultra-low power algorithm design required a deterministic, finite-state machine (FSM) to wake up from low-power mode, load the memory of the FSM, measure user activity and step counts, update memory, and return to low-power mode-for months on end.
• Wireless sensor (BLE) operates >6 months in the field using a single 240mAh coin-cell battery
• Leveraged Flutter software to generate a mobile, tablet, and web application using a single set of source code
• Link: OandP1

Riddell Axiom Product 2020-2021

• Designed the protocol to network 150+ wireless devices over long-range (>100m) using the BLE 5.0 protocol
• Impelmented a Cloud-based microservice used to calibrate the Axiom sensors remotely from 3 different manufacturing facilities
• Link: Axiom Helmet

Amulet Project 2014-2019 Link: Amulet Research and Development of Computational Jewlery for Mobile Health Applications.
Thayer School of Engineering at Dartmouth College, Hanover, NH

• Applying advanced digital signal processing techniques to enable long-term physiological monitoring on ultra-low resource microcontrollers
• Developed first long-term physiological recording device using a single 64kB microcontroller
• First to implement real-time compression of the electrodermal activity signal on an embedded system
• Instrumentation Design Experience: optical heart rate monitor (PPG), electrodermal activity sensor (EDA), 2-lead ECG heart rate monitor

The ICE-MITT Project Mobile Ice-Core Cooling System 6/2014
Thayer School of Engineering at Dartmouth College, Hanover, NH

• Designed, constructed, and implemented a mobile, 1 kW thermoelectric cooling system for the `transportation of ice cores from Barrow, AK to Hanover, NH.
• Developed cooling system was first to transport arctic sea ice at original temperatures.
• https://engineering.dartmouth.edu/magazine/dartmouth-engineers-develop-new-device-for-climate-change-research/

Hypertherm Sustainability Project 9/2013-3/2014 Thayer School of Engineering at Dartmouth College, Hanover, NH
21 Great Hollow Road, Hypertherm Inc., Lebanon, NH

• Worked in a team of 4 to assess the feasibility of a bio-fuel based commercial boiler to replace `a 2.2 MMBtu/hr diesel system.
• Provided automated calculator for alternative commercial heating systems using sustainable fuels.
• Presented findings and recommendations to board members (implemented: June 2015).
• https://engineering.dartmouth.edu/cook/previous/energy-environment

PUBLICATIONS

Taylor, Stephen, and Gunnar Pope. Hardware Sequence Combinators. International Conference on Cyber Warfare and Security. Vol. 19. No. 1. 2024. Mar 2024

Mishra, Varun & Pope, Gunnar & Lord, Sarah & Lewia, Stephanie & Lowens, Byron & Caine, Kelly & Sen, Sougata & Halter, Ryan & Kotz, David. (2020). Continuous Detection of Physiological Stress with Commodity Hardware. ACM Transactions on Computing for Healthcare. 1. 1-30. 10.1145/3361562. Jan 2020

Pope, Gunnar Crimmin. An Ultra-Low Resource System for Electrodermal Activity Monitoring. Diss. Dartmouth College, Jun 2019. PhD Thesis Link

G. Pope, V. Mishra, S. Lewia, B. Lowens, D. Kotz, S. Lord, and R. Halter, An ultra-low resource wearable eda sensor using wavelet compression, in 2018 IEEE 15th International Conference on Wearable and Implantable Body Sensor Networks (BSN), Mar 2018

Varun Mishra, Gunnar Pope, Sarah Lord, Stephanie Lewia, Byron Lowens, Kelly Caine, Sougata Sen, Ryan Halter, and David Kotz. The Case for a Commodity Hardware Solution for Stress Detection. In Workshop on Mental Health: Sensing & Intervention, pages 1717-1728, October 2018. ACM. DOI 10.1145/3267305.3267538 Oct 2018

PRESENTATIONS

An Ultra-Low Resource System for Electrodermal Activity Monitoring Jun 2018
G. Pope, Ph.D. Thesis Defense
Thayer School of Engineering at Dartmouth, Hanover, NH

An Ultra-Low Resource System for Electrodermal Activity Monitoring Jul 2018
G. Pope, Ph.D. Thesis Proposal Thayer School of Engineering at Dartmouth, Hanover, NH

Designing Wearables for Behavioral Health Research: An Engineer’s Perspective Apr 2017
Center for Behavior Health and Technologies, Lebanon, NH

An Ultra-Low Resource Wearable EDA Sensor Using Wavelet Compression Feb 2017
Engineering in Medicine Lecture
Thayer School of Engineering at Dartmouth, Hanover, NH

Challenges and Opportunities of Measuring Electrodermal Activity at the Wrist Oct 2016 Engineering in Medicine Lecture
Thayer School of Engineering at Dartmouth, Hanover, NH

Designing wearable, biomedical sensors for continuous stress assessment—a systems level approach Dec 2015 Engineering in Medicine Lecture
Thayer School of Engineering at Dartmouth, Hanover, NH

MEDIA

Dartmouth Team First to Transport Arctic Sea Ice at Original Temperature Apr 2015 A. Fiorentino, Dartmouth Engineer Magazine Link

Dartmouth Engineers Develop New Device for Climate Change Research Sep 2014 A. Fiorentino, Dartmouth Engineer Magazine, Link

VOLUNTEERING

XC Ski Coach for Ford Sayre, Hanover, NH Winter 2024
Mentor, EOS.IO Hackathon, San Fransisco, CA Nov 2018
Ski Patrol, Storres Hill, Lebanon, NH Winter 17/18
Ski Patrol, Storres Hill, Lebanon, NH Winter 16/17
Volunteer, Formula Hybrid, SAE Collegiate Competition May 2015
Volunteer, Formula Hybrid, SAE Collegiate Competition May 2014
Contest Judge, High School Science Fair, Windsor High School Jan. 2014
Contest Judge, FIRST Lego League March 2014
Volunteer, Formula Hybrid, SAE Collegiate Competition May 2013
Judge, FIRST Lego League March 2013

INTERESTS

Living Is Learning ~ Learning Is Living

Decision-making and risk, free climbing, powder skiing, mountain biking, paino