Keeling Labs increases the revenue of grid-scale batteries by optimizing how they buy and sell power with machine learning. We’re bringing the latest in reinforcement learning to a rapidly growing $2B+ market, backed by years of experience applying ML for batteries at Rivian.Backed by Y Combinator (W23)

- Started the ESS Data Science team in 2020, which grew to 9 members by my departure- Proposed, lead, and delivered a project to apply deep reinforcement learning to DC fast charging optimization. Demonstrated a 5m+/12% reduction in charge time on vehicle in final sim-to-real validation.- Created a highly scalable, cloud-based RL training system from scratch using containerized simulations on Kubernetes with a Spark-based analytics pipeline to monitor agent performance- Developed models to understand the performance of battery products in fleet data, particularly related to SOH- Built a near-real-time batch processing engine to automatically analyze and diagnose fleet data — generating event-specific KPIs and bespoke analysis/reporting related to the battery- Designed and built an end-to-end data pipeline to ingest, decode, and analyze terabytes of battery test data from research facilities in Normal, IL and Southern California

Algorithms & Controls Work: - Collaborated with the Cell Engineering Team, Thermal Management Team, and Energy Management Team to develop our charging controls on the BMS. Based on extensive cell testing and vehicle validation data, this control system determined the instantaneous current limit for the battery when charging (both DC Fast Charging and AC continuous charging) - Developed our isolation monitoring control system to measure the isolation resistance of the HV system - Created a unit-testing framework for model-based design using Simulink Test to validate our controls using MIL simulation - Worked with the algorithms team to define our state of health (SOH) strategy and initial implementation (continued in Battery Data Science Engineer Role)Data Science Work: - Designed and built a Rivian-internal Python package to conduct big data analysis on prototype vehicle data, pack testing logs, and lab data for the Energy Storage Systems team. This package is now used throughout the company for data science, primarily assisting with vehicle CAN log analysis and prototype-related big data analysis. - Created a cloud-based decoding pipeline for prototype vehicle data from scratch to make all vehicle logs query-able for big data analysis using S3, EC2, and the aforementioned Python package.- Conducted data analysis on numerous subsystems within the battery, our controls, and algorithms to assess performance and ensure requirements were being met.- Developing our offline state of health (SOH) framework and offline BMS data strategy (continued in Battery Data Science Engineer Role)

- Built a data analysis program in Python and SQL to analyze cell voltage imbalance in prototype vehicles, along with various other performance parameters- Designed a controls system for DC Fast Charging on the Battery Management System using Simulink and Stateflow- Determined the isolation resistance of battery modules by designing and carrying out numerous laboratory measurement techniques

- Conducted primary research on the upcoming CORSIA offset market (Carbon Offsetting and Reduction Scheme for International Aviation) and the capacity bidding process in the PJM Interconnection (Pennsylvania New Jersey Maryland)- Created a speculative Congestion Revenue Right trading strategy for the ERCOT market (Electric Reliability Council of Texas)- Analyzed project leases to calculate site lease payment run-outs- Learned about various aspects of project development and operation from team members of New Markets, Zoning, EPC, Environmental, Project Finance, and O&M during weekly intern meetings

- Researched solar economics, securities trading infrastructure, and blockchain technology to determine the implementation constraints and financial feasibility of a stock market for energy infrastructure (DIFEI)- Modeled a multi-axis-tracking photovoltaic solar farm using NREL’s System Advisory Model Software to demonstrate an asset on the framework’s equity exchange

- Co-founded a mobile app company that launched two apps (Packup & Lookabout)- Managed a team of three, was responsible for product design, business partnerships, 
fundraising, and distribution- Analyzed usage data in Flurry and AppBoy to make data-driven product rollout decisions- Lookabout (360° photo-sharing) got over 1,000 users in more than 10 countries- Investment from Science Inc. (Dollar Shave Club, DogVacay, Famebit)------You can see more on Lookabout here: www.lookabout.io

- Developed a Matlab data visualization tool to expedite data analysis for the group’s development of a nanoscale magnetic sensor- Created a multivariate chi-squared estimation algorithm in Matlab to solve for sensor parameters using free-induction decay and spin-echo data- Assisted three graduate students in lab operations and tested various equipment- Awarded the USC Dornsife Summer Undergraduate Research Fund (SURF) grant