Haga

Data room

External investor sharing

Team

Hire / co-founder scorecard

Ready

Written criteria for first hire or co-founder — ML systems / robotics eval; quiet search, no placeholder equity.

Locked decision: Haga is solo-founded by Mushood Hanif. Team strength comes from an active, criteria-driven hire / co-founder search — not a multi-name roster for optics.

Cap table at incorporation: 90% founder / 10% option pool (locked within the 85–90 / 10–15 band). Candidates join via new grants from that pool (or a negotiated co-founder grant when someone clears). Do not reserve named equity for people who have not committed.


When outreach starts

Trigger Action
Evidence pack Ready (Decision 0) Quiet outreach may begin
Soft raise / post-wire Full-time hire capacity increases (compute + runway)
Candidate clears scorecard Structure as employee (standard grant) or co-founder (larger grant + vesting) — only after mutual commit

ICP buyer conversations and design-partner evals do not wait on a hire.


Role shape (pick one track per candidate)

Track Title posture When to use
First hire ML / eval engineer (or equivalent) Strong builder who wants employment + equity; founder remains sole founder
Co-founder Co-founder, ML systems / robotics eval Peer-level ownership of harness, scoring, and technical GTM; clears bar below and wants founding risk

Do not invent intermediate titles to look larger. One sharp hire beats five soft “founders.”


Must-haves (all tracks)

Candidate must clear all of:

  1. ML systems depth — ships models or evaluation pipelines in production or research-grade code (not demos-only)
  2. Robotics or physical-AI adjacency — policy eval, sim, world models, or closely related (CV / control / synthetic data with physics awareness)
  3. Reproducibility instinct — seeds, configs, CIs, failure cases; comfort defending numbers in diligence
  4. Async ownership — can own a pillar (Pillar 1 or 2) or continuous-scoring design without daily pair-programming
  5. Values fit — independent-auditor posture; no urge to turn Haga into a training-loop or public leaderboard product

Strong signals

  • Prior work on robotics benchmarks, sim randomization, physics checkers, or world-model QA
  • Published or open reproducible eval code with honest negative results
  • Experience turning research harnesses into customer-facing reports or APIs
  • Comfort talking to ICP (policy labs, synthetic-data vendors) without overselling

Disqualifiers

  • Wants co-founder title without full-time commitment or IP assignment
  • Primary interest is training / Sim2Real product (wrong company)
  • Cannot explain a past eval result with seeds, n, and claim boundaries
  • Expects reserved equity before contribution or before incorporation hygiene

Deal structure (draft)

Element Draft
Equity source Unissued option pool (10% FD at incorporation) — new grants only
Employee grant Standard early-hire band; 4yr / 1yr cliff unless counsel changes
Co-founder grant Larger slice from pool or negotiated founder stock if joining at / before incorporation — counsel + sole founder approve in writing
IP Assignment into Delaware entity (or Gate A side letter pre-Atlas) required before meaningful code access
Cash Competitive for stage once raise closes; pre-raise may be equity-heavy or grant/contract

Exact percentages stay internal until a named candidate is in diligence — not a public rate card.


  1. Score against this page (must-haves first)
  2. Technical deep-dive: critique a Haga Pillar 1 or Pillar 2 result against the public Lab and methodology
  3. Working session: scoped harness or report improvement (paid trial if needed)
  4. Reference check on shipping + honesty under ambiguity
  5. Written offer: track (hire vs co-founder), grant, vesting, IP — then counsel templates

Log serious candidates privately (not in git). Public site and pitch stay sole founder until someone has signed.


Bus-factor mitigation (what investors hear)

Solo founder is acknowledged. Mitigation is this scorecard + active search + option pool reserved for builders — not a fake co-founder list. Post-close milestone M4 (locked): first ML/eval hire or co-founder offer accepted within 9 months of wire — Milestone model · Use of funds.

Related: Bios · Founder equity & IP · Use of funds.