ML Engineer — datasets tooling & open data (GitHub: London)
Daniel van Strien is a Hugging Face ML engineer working on datasets tooling and open-data infrastructure (public GitHub location: London) — behind the pipelines that turn raw public corpora into Hub-ready datasets. NVIDIA agreed to acquire Hugging Face for ~$12.9B on 2026-09-03 ($11.9B to shareholders + up to $1B equity retention for HF employees joining NVIDIA; HF to remain an open platform; expected close H1 2027 pending regulatory approvals; antitrust runway + GPU-neutrality of the Hub are the live radar questions). Listed in Hugging Face's public organization directory (187 members, retrieved 2026-09-04); named at the HF All Hands where the deal was announced.
Only public nodes listed.
1 entities geo-pinned · toggle People / Institutions · 51.5074, -0.1278
🔭 WATCH: whether the rail survives its own token · rail volume, not follower count · which chain the settlement moves to next
🔍 Daniel van Strien is carried at MED confidence on 3 filed sources. The horizon below is a hypothesis with a shelf life, not advice.
Paste the badge on a site, a README or an agent page. It serves live from the edge and links back with its own attribution code.
Machine form: GET /badge/daniel-van-strien.json
Paste a Base wallet and this record joins your watchlist. Nothing is uploaded, nothing is stored off-chain beyond the list, and removing it is one click.
Paste the badge on a post, a deck or a partner page. The image links home with ?ref=badge-daniel-van-strien, so every embed is attributed in the referral ledger and every reader lands on this record.
JSON twin: /badge/daniel-van-strien.json returns the same snippet for a generator.
Paste the badge on a post, a deck or a partner page. The image links home with ?ref=badge-daniel-van-strien, so every embed is attributed in the referral ledger and every reader lands on this record.
JSON twin: /badge/daniel-van-strien.json returns the same snippet for a generator.
Paste the badge on your own page. The image is edge-cached SVG, the link carries the attribution code, and the counter moves in the referral ledger.
The record is what it is: ML Engineer — datasets tooling & open data (GitHub: London). Nothing below is filed — it is the shape the seat would take if the current trajectory holds, and it is re-scored every time a new public source lands.
🔭 whether the rail survives its own tokenExpect the mandate to widen before it deepens. Watch whether the current seat adds an adjacent function to the seat rather than a bigger title on the same one.
🔭 rail volume, not follower countA second mandate is the normal move here — a board seat, a fund-side advisory role, or an operating chair next to the current one. The tell is a first public co-signature, not an announcement.
🔭 which chain the settlement moves to nextFirst capital event window. Either a priced round behind the current seat or a personal vehicle appears. Watch which chain the settlement moves to next rather than the headline valuation.
🔭 fee take-rate disclosureJurisdictional switch is the highest-probability structural change from London, UK. A second residence, a second entity, or a second currency book (USD to a settlement unit) is the observable.
🔭 agent-side integration countInstitutional shell forms: a holding vehicle, a family-office wrapper or a named co-investment club. Shells are quiet — the disclosure arrives through registry filings, not posts.
🔭 whether the rail survives its own tokenDistribution becomes the constraint. At this horizon the question is who carries the current seat's output and this seat's name, and on what terms. Watch fee take-rate disclosure.
🔭 rail volume, not follower countSuccession pressure. Either a named deputy appears under the seat or the seat is re-scoped away from day-to-day. Both read as the same signal from outside.
🔭 which chain the settlement moves to nextSecond-order liquidity: secondary sales, a tender, or a strategic minority into the current seat. This is where an OSINT file usually goes stale for six months and then re-prices.
🔭 fee take-rate disclosureRecord consolidation. By here the file should be readable as a single arc rather than a list of events, and the London, UK base stops being the interesting part.
🔭 agent-side integration countEnd-of-horizon check. Re-read 2026 against 2036: the useful output is which of the eleven rows actually printed, and which were noise carried by the market rather than the person.
🔭 whether the rail survives its own token🔍 Deterministic horizon model: every row is derived from facts already on this record and re-scored when a new public source lands. No forecast value is asserted.
Buy without a browser: GET /oauth/protected-resource, then tools/call fetch_profile over MCP. Bulk and 402 retries are idempotent on the same slug.
Machine pack: /api/um-radar/agent-pack?slug=daniel-van-strienBatch / dataset pricing on request
The dossier prose stays public. The space around it is for sale: locales, frames, video, links, the city slot, the verified badge. Holding $UM cuts the price at quote time - the balance is read from your wallet, nothing is self-declared.
Public OSINT radar only. Not investment advice. Not a prediction engine.
Creator-fee revenue is onchain and verifiable.