← The Independent 100: a study

Varun

#19 in The Frontier Outside, the curator’s rank and section @varun_mathur Relevance to our work: 3 of 5

Bracketed tags are evidence labels: [Verified] means we read the page, file or API response ourselves, [Reported] means a source says so and we did not check the underlying thing, and [Inference] is our own reasoning. CI is continuous integration: automated checks that run on every push.

Credit

The Independent 100 list was curated by dan (@irl_danB) and published on 2026-09-24 at independent.prose.md (announcement). The ranks and section names are his, and number 100 is his open wild-card slot. This study is ours and is not endorsed by him or by anyone on the list.

What they publicly do

Founder and CEO of HyperspaceAI [Reported: X profile x.com/varun_mathur]. Hyperspace publishes a P2P AI inference network client, an 'Agentic OS' of 26 sidecar services, and a repository of agent-run distributed 'autoresearch' experiments [Verified changelog.hyper.space, github.com/hyperspaceai/agi; figures are self-reported]. His recent posts describe using Hyperspace with TypeSafe's Jev model and jevcache [Verified x.com/varun_mathur/status/2101740594754093349].

Public work

  • hyperspaceai/agi repo MIT last activity 2026-09-17 · 2,062 GitHub stars on 2026-09-24

    Living research repo written by agents that run Karpathy-style autoresearch loops and share results over P2P gossip [Verified repo description; results Reported venturebeat.com/technology/andrej-karpathys-new-open-source-autoresearch-lets-you-run-hundreds-of-ai].

  • jevcache repo no license file last activity 2026-09-19 · 72 GitHub stars on 2026-09-24

    Decision cache/ledger for Jev-class typed-judgment models: redacts and canonicalizes state before hashing, caches decisions locally, replays fixtures in CI to catch drift; distribution repo only (README + image), Rust binary via curl|sh [Verified README and contents listing].

  • Agentic OS changelog (AVM security gate) product no license stated last activity 2026-04-30

    Shiplog of 26 sidecars; AVM gate described as 'rules first, ML advisory: ML can restrict, ML cannot relax, block always wins' (2026-04-28) [Verified changelog text].

  • hyperspace-node repo no license file last activity 2026-07-30 · 329 GitHub stars on 2026-09-24

    Client distribution for the P2P inference network (CLI, tray app, browser) [Verified repo description; node counts self-reported].

  • How I invented gossiping agent swarms blog no license stated last activity 2026-09

    His account of orchestrator-free agent swarms [Reported via search; not read].

Licenses of GitHub repositories are from the authenticated GitHub license API on 2026-09-25, one row per repository in evidence/licenses.jsonl (“no license file” means the API found none; “license file present, not recognised by GitHub” means GitHub could not classify it). Other licenses are as the source states. Check the license file before reusing anything.

Relevance to our work

3 of 5 lenses: agents, security, memory, orchestration, local-models

jevcache targets TypeSafe's Jev model, and its redact-then-hash key and CI replay are ideas any client for that model could apply [Inference]. The AVM rule 'ML can restrict, never relax' is a crisp doctrine for agent hooks and guards [Inference]. Most shipped code is unlicensed binaries, so this is ideas, not adoption.

Scored 0 to 5 against the study’s relevance rubric: 5 means public, openly licensed work squarely in our lanes that we could study or adopt now, and 1 means little connection to our work. The score is one reviewer’s judgement of how much this public work bears on our own work. It is not a ranking of people and says nothing about the person or the quality of their work.

What we would study next

  1. jevcache README design (github.com/hyperspaceai/jevcache). Its fingerprinting (redact PII and volatile fields before hashing, per-schema salt) and replay mode are a compact design for caching and replaying model decisions. Effort: small.
  2. AVM entries in the changelog (changelog.hyper.space). Rules-first, ML-advisory gate design with an adversarial corpus; relevant to prompt-injection guards. Effort: small.
  3. hyperspaceai/agi repo layout (github.com/hyperspaceai/agi). How agent-written experiment results are recorded and peer-scored; relevant to any evidence ledger. Effort: medium.

Before reusing anything: jevcache, hyperspace-node and aios-cli have no license file (default all rights reserved) and install closed binaries via curl|sh; do not adopt. agi is MIT. The platform includes blockchain, wallet and points components outside our scope. Hyperspace also ships a product called 'Agent Mail' (agent mailboxes) that is unrelated to MCP Agent Mail; watch for name confusion. Network size, download and experiment figures are self-reported.

Sources

Overall confidence in this record: Medium. Sources were read on 2026-09-24 and every link was checked on 2026-09-25. The full record is line 19 of people.jsonl.

Limits

Not verified: Network size (2M+ nodes), download counts, experiment and trading results; jevcache source code (not published); whether AVM source is public; his Substack post body.

This record comes from a first-pass survey of public material by an AI agent session on 2026-09-24, edited for publication on 2026-09-25. Nothing was installed or run for it. Roles, stars and licenses change; see the study’s limits.

Request a correction

Is something about this record wrong or out of date, or would you rather not be covered? Open a correction (a GitHub issue form with this page’s address filled in). A request to remove a record is honoured without argument; accepted corrections are recorded in the changelog with the date.

Help fix it

Each opens a GitHub issue form with this page’s address filled in. You need a GitHub account.

Made by Joshua Nowak (ZestStream) with AI coding agents. Not affiliated with or funded by Jeffrey Emanuel; Joshua uses his public tools and is a paying subscriber to jeffreys-skills.md.

Source on GitHub MIT licensed