(stealth)
Diego Santos (@dsantos_ml) · Undisclosed — post-training data curation at scale.
Every claim in this memo, checked against how well it holds up.
Company Snapshot
What the company does and where it sits in the market.
(stealth) operates in AI Infra (Lisbon, PT, Idea). Undisclosed — post-training data curation at scale. Post-training data curation is hot and under-tooled.
Investment Hypotheses
The core bets underlying this memo, each traced to its evidence.
- Team quality: First-author on 3 cited papers, now defecting from academia (affiliation → gmail). Repo shows productization. High latent readiness; not yet activated.⟦ev-ds-arxiv⟧
- Market wedge: Post-training data curation is hot and under-tooled.
- Defensibility: Thesis-shaped; needs the founder to actually commit. Activation pending.
- Founder credibility: First-author, 3 cited papers⟦ev-ds-arxiv⟧VERIFIED
SWOT
Strengths, weaknesses, opportunities, threats — shown only where evidence exists.
- First-author, 3 cited papers⟦ev-ds-arxiv⟧VERIFIED
- Founder axis improving: First-author on 3 cited papers, now defecting from academia (affiliation → gmail). Repo shows productization. High latent readiness; not yet activated.⟦ev-ds-arxiv⟧
- Market axis improving (bullish): Post-training data curation is hot and under-tooled.
- arXiv Affiliation Defection: Institutional affiliation replaced by a personal gmail — leaving academia with his tech.
Problem & Product
The problem being solved and how the product addresses it.
Problem: Post-training data curation is hot and under-tooled.
Product: Undisclosed — post-training data curation at scale. Thesis-shaped; needs the founder to actually commit. Activation pending.
Traction & KPIs
Traction and revenue signals, with unverified gaps flagged explicitly.
- ARR: not applicable — pre-revenue stage
- Cap table: not disclosed
- Financials / P&L: unavailable at this stage
1 sourcebehind the citations above — duplicates the inline ⟦refs⟧, expand to inspect the raw excerpt.
Three-Axis Verdict
never averagedFounder, market, and idea-fit scored independently — one weak axis can't be masked by a strong one.
Highest-latency, highest-upside: activate with cold outreach to trigger a real application, then run the Arena.