Projects we've worked on

Listening at Scale

Building an audio pipeline to monitor biodiversity through sound.

Rather than listening to hours of nature recordings, we built a system that reads the audio first: flagging quality issues, running an initial analysis, and preparing the data before any species classification happens. From there, the system runs recordings through multiple machine learning models, compares their results side by side, and fine-tunes confidence scores so the client can trust what the models are telling them. What used to be a manual listening exercise is now a repeatable, scalable process for tracking biodiversity change.

DATA & TECHNOLOGY | OPERATIONS

Designing for Growth and Trust

A blueprint to help a fast-growing registry scale on solid ground.

A customer that outgrew ad-hoc coordination needed a clearer way to work: less manual data entry, fewer bottlenecks in review and approval, and a system built for multiple stakeholders. We ran a design-thinking sprint to map the full workflow, find where time and transparency were being lost, and design a future-state process that kept the rigor while removing the friction. We tested it with a working pilot before scoping the full system now being built, one that leverages AI to handle the manual work and keep the process running at scale.

STRATEGY | DATA & TECHNOLOGY | OPERATIONS

From Protocol to Proof

Helping a nature-based carbon startup earn its first issuance.

A nature-based carbon capture startup needed to navigate a system that rewards precision: which protocol to pursue, how to quantify emissions, and how to build an application that could hold up to close review. We helped shape their MRV strategy and emissions quantification approach, guiding the process from protocol selection through to submission while keeping their scientific expertise front and center. The result: their first batch of high-integrity carbon credits

CARBON PROGRAM EXECUTION | DATA & TECHNOLOGY

Projects we've worked on

Listening at Scale

Designing for Growth and Trust

From Protocol to Proof

Have a problem that looks something like this? Let's talk about it.