Zentri Research
Apply →
Cohort 01 · Timeline in progress · Fully remote

A public research lab
for ambitious AI builders.

You bring the research. We handle everything else.

Apply to Cohort 01 → What we handle
Cohort 01Open
ApplicationsIn progress
Tracks live05
Start dateTBD
Portrait · 2026 Iker Moel Tacher — Founder & Director, Zentri Research
FIG. CEO Remote
Founder & Director

Iker Moel Tacher

“I built Zentri because I watched too many capable researchers sit outside institutional gates.”

No collaborators, no compute, no place to publish. Zentri is the lab I wished existed when I started: public by default, built to help ambitious researchers ship serious work without waiting for institutional permission.

Iker M. T. Talk to Iker →
§ 02 — What we handle

You focus on the research.
We handle the rest.

01
Funding
Microgrants, compute credits, and milestone bounties. You never write a grant proposal.
We secure it
02
Marketing & distribution
Launch posts, newsletter, X threads, research channels. Your work gets read.
We amplify it
03
Team management
2–6 serious collaborators, weekly progress, blockers cleared. You show up and build.
We organize it
04
Publishing
Editorial pass, then arXiv + the Zentri Research journal with a permanent DOI.
We publish it

Authorship stays with you. Always.

How funding flows to your team.
Live · Cohort 01
Sponsors In progress Grants Milestone-based Donations Public Compute partners In-kind Zentri Pool MERIT-BASED ALLOCATION Team support By request Compute credits GPU hours Milestone bounty On delivery Publishing & DOI Covered Distribution / PR Handled INPUTS POOL TO TEAMS
Source / Destination Capital flow Active allocation
Pooled funding In progress Cohort 01 · numbers publish at launch
Microgrant range By request Per team · structure in progress
Compute sponsored In progress Partner commitments underway
Takes from teams 0% No equity · no IP · ever
§ 01 — Mission

Consequential AI research, in public.

For the talent outside the gates.

A · Gated researchCurrent state

Talent clusters inside a few labs.

WALL ✕ ✕ ✕

Capable researchers stay outside. Work never happens, or happens alone without collaborators, compute, or a place to publish.

→ Zentri →
B · Public researchWhat we build

A shared, open lab for everyone else.

LAB

Distributed teams, central infrastructure. Merit-selected, publicly funded, fully open output. Structure where institutions offer none.

Purpose

Genuinely understand the world.

Build AI systems that go beyond next-token prediction — architectures with causal reasoning, persistent identity, and real agency. Not tools. Intelligence.

Mission

Organize the talent outside the gates.

Structured sprints, small teams, merit-based microgrants. We give serious researchers what institutions give theirs: collaborators, structure, accountability, and funding.

Vision

Research as public infrastructure.

Every experiment, every paper, every dataset — open. A lab whose outputs belong to everyone, not a quarterly earnings report.

§ 03 — Research Tracks

Five tracks. One discipline.

T · 01
Architectural Innovation
Beyond Transformers. State-space, recurrence, causal structure.
Accepting
T · 02
Transformer Optimization
Attention, throughput, inference, quantization.
Accepting
T · 03
AI Agents
Planning, tools, long memory, real-world action.
Accepting
T · 04
Interpretability
Circuits, probes, feature analysis.
Accepting
T · 05
Experimental AI Ideas
The weird stuff. Too strange for a standard lab.
Limited seats
§ 04 — The Six Weeks

Idea to public output, timeline in progress.

Week 01
Proposal
One-page thesis, posted publicly by Friday.
Week 02
Setup
Data, baselines, compute, repo.
Week 03
Build
First runs. Does the thesis survive reality?
Week 04
Iterate
Ablations, stronger baselines, or an honest pivot.
Week 05
Test
Freeze. Full eval. Peer review.
Week 06
Publish
Demo day. Repo. Paper draft.
§ 05 — Structure

How the lab runs.

The numbers

Cohort size
In progress, published before launch.
Teams
Formed per project, final size TBD.
Cycle length
6 weeks, no extensions.
Tracks open
Five, plus experimental.
Microgrants
Merit-based, range in progress.
Compute
Sponsored credits for selected teams.

Non-negotiables

  1. 01Weekly public progress update. No post, no place.
  2. 02Every output ships public — repo, writeup, data.
  3. 03Teams failing their thesis pivot or dissolve. No zombies.
  4. 04Inactivity for 7 days = out. We protect the remaining team.
  5. 05Peer review is mandatory, kind, and technical.
  6. 06No pay-to-enter. Merit only. Microgrants flow toward work, not membership.
§ 06 — What teams ship

Every cycle ends with real artifacts.

01
Paper draft
arXiv-ready writeup.
02
Public repo
Reproducible, one-command run.
03
Live demo
Interactive, on the web.
04
Benchmark
Numbers that can be cited.
05
Technical writeup
For the parts that aren't paper-shaped.
§ 07 — The lab, internally

A tiny operation. Intentionally.

“A machine for accelerating researchers who would otherwise never meet a lab.”

— Cohort 01

  • DirectorSets the tracks, selects the cohort, owns final calls on funding and removals.
  • ReviewersVolunteer senior researchers who peer-review weekly updates and final outputs.
  • TeamsSelf-organized. A team lead is declared by week one and is accountable to the cohort.
  • FundingSponsor-funded microgrants, compute credits, and milestone bounties. No pay-to-play.
  • PublishingSubmission guidance and endorsement support for eligible work. Authorship stays with the authors.
  • GovernanceTransparent. Selection criteria, funding decisions, and removals are logged publicly.
§ 08 — Apply · Cohort 01

Bring one real idea. Leave with shipped research.

One button. Five minutes. 10-day response.

Apply now → Sponsor a track
Cohort 01 · 10 seats reserved for underrepresented regions