Caleb John

I'm a founder and engineer who works across SaaS, heavy industry, energy, and research. I build technology for complex, ambiguous problems.

Co-founder — LucidBox

May 2025 – present
A specialty engineering firm in Alberta that helps startups build their initial technology, with clients ranging from SaaS software to heavy industry and energy technology, from seed stage through later funding rounds.

Co-founded the firm; secured our first four clients and achieved profitability in our first year.
Lead technology decisions and initial builds for clients across various domains.
Translate complex, ambiguous problems into practical technical solutions.

First Engineer & Engineering Lead — Solartility

Sep 2022 – May 2025
Technology to intelligently manage power and energy flows using distributed energy resources (DERs) and virtual power plant (VPP) technology.

First engineering hire; grew and led the engineering team from just me to five.
Built a real-time energy-storage arbitrage engine that doubled the year-over-year value of a customer's solar system relative to panels alone on a standard tariff, combining smart control with modern tariff design.
Led development of the residential VPP product and drove its technical and architectural direction.
Designed and initiated a DERMS built on Postgres, and architected the AWS cloud deployment for the product backend.
Ran the numeric analysis and built the demo application behind a fundraising effort that secured over $5M in grants.
Set engineering security guidelines using OWASP for a defence-in-depth posture.

With the platform now in place and running, I used that experience to co-found LucidBox.

Algorithm Engineer — Energy Toolbase

Apr 2021 – Jul 2022
A suite of project modeling, control, and monitoring tools for solar-and-storage developers; the core product is algorithmic control of large-scale batteries.

Developed a legacy Python battery simulator into a core product over nine months.
Reduced simulation runtime to one third of the previous time through iterative profiling and optimization.
Led the cross-departmental AWS deployment of the energy-storage-system (ESS) simulator.
Established engineering processes that let the monorepo scale sustainably; initiated and led a technical study group.

Firmware Engineer — Vermillion

Mar 2019 – Nov 2021 · contract, alongside other roles
A Calgary power-electronics design shop building battery-storage inverters that scale from very small to mid/large systems.

Assumed development of non-functional lighting-inverter firmware, subsystems that worked in isolation but failed together, written by a team of six over a year. Restored intended functionality within two weeks and exceeded original performance goals on real hardware within six weeks.
Wrote firmware to control the submodules of a battery-storage inverter, and performed electronic verification and debugging.

Machine Learning Research — M.Sc.

Sep 2018 – Apr 2021
Built a system to generate human-interpretable predictors of chronic homelessness from standard homeless-shelter data.

Designed and built an interpretable rule-learning pipeline — the RuGRATS system — using industry-grade ML practices such as cross-validation.
Trained and evaluated on 12 years of service interactions covering over 40,000 individuals from a major North American shelter.
Implemented efficient classification-rule-learning algorithms in numpy and pandas.
Published the results in several peer-reviewed papers (below).

Photonics Research — NTT BRL

May 2016 – April 2017
Researched photonic nanostructures, simulating and building experimental hardware at the boundary of optics and quantum physics.

Designed, simulated, and tested novel microwire-laser structures.
Simulated and verified photonic-crystal nanostructures using internal FDTD software and commercial (COMSOL) FEM software.
Worked in-lab to identify quantum dots on photonic substrates.

Quantum Communications Research — University of Calgary / TU Delft

Summers 2014, 2015, 2017, 2018
Built experimental quantum communication hardware.

Designed and implemented qubit-preparation and QKD control systems on FPGAs (VHDL) — generating pulses under 1 ns wide with under 5 ns spacing — along with the GUI control software.
Built a MATLAB pipeline to extract features from tens of GBs of raw time-series data from a quantum random number generator and prepare it for statistical testing.

Open Source

Joplin (cross-platform note-taking app; Electron/React/Redux): three-time Google Summer of Code mentor. Implemented color themes and the initial spell-checking, replaced the core markdown editor with a more capable one, and built several widely-used plugins.