JOSH STEVENSON / AI SYSTEMS ENGINEER
I build agent runtimes, durable memory, and local AI infrastructure. Open source. Clear boundaries. Evidence you can inspect.
NVIDIA PAIR contributor Merged upstream · September 2026CONTRIBUTIONS & RECOGNITION
Accepted upstream code and public recognition from the agent community. Follow the original sources.
ACCEPTED UPSTREAM CONTRIBUTION
My cluster-manager fix was accepted into NVIDIA’s Personal AI Router. Trust-change notifications now follow successfully persisted endorsement changes, with regression coverage for duplicates and failed writes.
An accepted open-source contribution to NVIDIA PAIR. This does not imply employment, sponsorship, or a commercial partnership with NVIDIA.
COMMUNITY RECOGNITION
Teknium, creator of Hermes Agent, highlighted my RecursiveIntell-enhanced Hermes demonstration and later reposted my work. That work has grown into Ares, my downstream agent workbench.
This is a public interaction around Josh’s engineering work, not a customer testimonial, partnership, or product endorsement.
01 / SELECTED ENGINEERING
Three connected areas of work: agent execution, persistent memory, and evidence. Every project links to source and names its current scope.
A Hermes-compatible agent workbench with managed runtime releases, rollback, and explicit evidence boundaries.
Source reviewed October 1, 2026 at Ares revision fadd47d7330445969cd99863e736c9b5dbfced54. The repository documents inherited Hermes surfaces, Ares-owned lifecycle code, gated execution, and external services separately. The September integration case remains available as a dated historical snapshot.
Ares is an independent RecursiveIntell downstream distribution, not an official Nous Research product. Source presence does not establish installed activation, every provider/platform combination, or production fitness.
Persistent agent memory with source history, temporal retrieval, and explicit device ownership.
Public repositories, crates, documentation, tests, and a dedicated product/proof surface.
Public artifacts establish implementation scope, not customer adoption, universal correctness, or production fitness.
Typed execution graphs and claim/evidence history that keep observed outcomes separate from asserted success.
Published crates, public source, schemas, examples, and test surfaces that can be inspected directly.
Receipts report observed execution. They do not prove factual truth, authorization, security, or task success.
02 / THE MEMORY SYSTEM
A personal, self-hosted agent memory server. Keep durable context, source history, and device ownership close to the system that uses them.
Explore Mnemes ↗Public software and research. Hardware deployment and production fitness have their own evidence requirements.
03 / WORK WITH ME
Bring a repeated workflow, a knowledge problem, or an agent system that needs a more reliable foundation. We’ll define a focused engagement.
A repeated process takes too much time. You need to decide which parts are useful to automate.
02 / Bounded implementationOne agent-assisted workflow has a clear owner and can be tested without reorganizing the whole business.
03 / Source-aware systemUseful answers are trapped across documents, notes, policies, or specialist knowledge that people repeatedly search for.
04 / Technical advisory or operationsYou need architecture judgment, reliability work, local-first deployment, Hermes customization, or continued improvement after a pilot.
Hiring for agent infrastructure? Working through a difficult system? Tell me what you’re trying to make work.
Start a conversation ↗