This is what's live in the lab right now and recent case studies that we've delivered.
Solid State Pros runs an open research program alongside our client work: original R&D into how computation, hardware, and learning systems can be built more visibly, more simply, and closer to physical reality. Each project below links to the full write-up on greenforest.io. Below the research, you'll find case studies from our enterprise practice, proof that the same thinking ships at global scale.
Maker-scale active-device boundary research
Makers can cut, print, mill, and etch almost anything, except the active devices at the heart of electronics. This research maps exactly where that boundary sits and what it would take to bring transistor-level fabrication down to the workbench, so builders aren't permanently locked out of the most important layer of their own tools.
One-pin radio and minimum radio mechanism
How much circuitry does a working radio actually require? This project strips wireless communication down to its bare mechanism, one pin at a time, to find the true minimum. It's both a serious question about what radio fundamentally is and a path to radically simpler, cheaper wireless for tiny systems.
Tiny learning machines that still actually learn
Frontier AI keeps scaling up. We're going the other direction: how small can a transformer get and still genuinely learn? Our four-layer tiny-transformer training runs probe the floor of machine learning, aiming at models small enough to understand completely, and someday small enough to live in hardware you can hold.
Cartilage and visibile reconfigurable computation
Cartilage is our long-horizon answer to a question we've chased for two decades: how do you grow computing systems without hidden complexity? It's a reconfigurable substrate where structure, ownership, and behavior stay visible as systems scale, computation you can see, reshape, and reason about, demonstrated through working nested-instantiation prototypes.
MUX tile / physical logic-building systems
Six snap-together tile types, built on a single primitive, are enough to construct any digital logic on a tabletop. The MUX tile system lets kids and adults physically build the inner engines of interactive systems, see causality directly, and scale their own designs, no sealed chips, no black boxes. This is the bridge product on the path to Cartilage.
Case Study: Agentic AI for Legal Compliance — Global Technology Leader
A Fortune 50 technology company needed to automate high-volume legal-compliance casework that consumed hours of expert time per case. Steve led the AI initiative applying LLMs, OCR, RAG, and agentic workflows to the pipeline, collapsing multi-hour case processing into minutes and proving that agentic AI can carry real regulated work, not just demos.
Case Study: Next-Gen Digital Commerce — lululemon
As Senior TPM for lululemon's Next Generation Commerce platform, Steve led delivery across a 40-person organization, 80 APIs, and 6 SAFe Scrum teams, modernizing the digital backbone of a $13B retailer with 600+ stores while the business kept running at full speed.
Case Study: Global Platform Transformation — Ingram Micro / CloudBlue
Steve co-directed a $300M digital transformation spanning 35 countries and a 700-person Agile team across 16 time zones, and designed GMOT, a global marketing orchestration tool integrating Marketo, HubSpot, Sprinklr, and Meltwater with the B2B2B commerce platform. Strategy to execution to adoption, at the largest scale there is.
"Looking to bring us onto your team? → Hire Us.
Want to back the open research with no strings? → Greenforest R&D Fund.