Learning, memory & adaptation
Exploring cognitive computation and the relationship between experience, prediction, and action. Interactive tasks offer a way to ask specific questions about an experimental system’s behavior.
Independent research · AI-assisted engineering
Stay curious.
Build. Test.
Ask again.
I’m Sterling Tuttle. I’m interested in how systems learn, remember, and adapt—and how that work can help us build a more sustainable future.
01 / Research interests
Big questions.
Concrete testing grounds.
Simulation & control
Duck simulations and adaptive-control experiments provide a tangible setting for studying movement, feedback, and failure. Simulation is a testing ground for questions about embodied systems.
Prediction & evaluation
Investigating signals, comparisons, and the conditions under which a prediction is useful. The challenge is to distinguish an interesting historical result from an advantage that survives new tests.
The work is experimental. I use interactive tasks, simulated environments, and historical prediction tests to examine specific questions. Published results will identify the version tested, comparisons, and limits of the evidence.
02 / Work you can inspect
From the question
to the evidence.
Public project · In development
TinyHive: coordinating AI agents
A public software project exploring how agents communicate and coordinate work. The repository is a place to inspect the implementation, follow its development, and ask concrete questions about its behavior.
Engineering review · November 2026
Making message contracts explicit
A recent AI-assisted review checked TinyHive’s payload handling across four retrieval methods. Twelve fixture cases behaved consistently; an older raw-string expectation differed from the returned JSON envelopes. The report separates that component check from full-system testing and records what still needs a decision.
Community contribution · November 2026
Reviewing evaluation tooling
An AI-produced follow-up review on Prime Intellect’s Verifiers project checked a proposed prompt-rendering change and suggested a regression test for preserving literal placeholder text. Five isolated function-level checks passed. The public comment credits the original author and separates compatibility checks from repository-wide testing or model-defense effectiveness.
I guide research direction and experimental decisions; AI does most coding. The maintenance review above is newly performed AI-assisted work, not a claim that I personally wrote a historical repair. Project source and component checks establish specific evidence, not production reliability.
03 / The person behind the work
James Sterling Tuttle.
Also known as
Jim James & V.
My background spans partner and affiliate systems, paid search, web development, and business operations. Alongside that work, I pursue independent research into learning, memory, and adaptive systems. I learn by building, testing, and staying with the questions.
I originate ideas and research directions, then lead AI teams through implementation and testing. AI does most of the coding. My role is to frame the experiments, challenge the results, and decide what to investigate next.
I’m working toward a research-lab role where I can collaborate with curious people on difficult problems. I’m also open to AI engineering and consulting that builds useful experience and moves that work forward.
04 / Working principles
The next question
has to earn its place.
- 01Make the question testable.
Define the observation and the comparison before deciding what a result means.
- 02Keep the failures in view.
Corrections and negative results help determine which direction deserves more work.
- 03Connect ambition to useful problems.
A sustainable society shaped by AI is the larger goal. Problems we face today are where the ideas must prove useful.
05 / Research · Engineering · Collaboration
A good question is
a good place to start.
Working on something difficult? I’m interested in research opportunities, substantive technical conversations, and AI engineering projects where I can contribute and learn.
sterling@sterlingtuttle.com ↗