TL;DR. We map the USCIS extraordinary-ability checklist — the criteria an O-1A or EB-1A case is built on — onto a 1.17-billion-profile graph, then rank and contact the plausible candidates. The strongest finding: intent and recent change beat a static "strong profile" — someone who just relocated, got promoted or defended a PhD is a warm lead reachable at the decision moment, worth more than a stronger-but-static name. The honesty boundary is fixed: tiers are evidence-strength for an attorney to qualify, never eligibility verdicts — USCIS makes the final call, our data cannot.
01What the data can see
Before any headcount, the honest question: which USCIS criteria can our data actually see? Each of the 8 O-1A criteria is graded below — strong, partial, weak, or blind.
| USCIS # | USCIS criterion | Our data source | Coverage | How we use it |
|---|
| 7 | Leading / critical role | career-position graph + company prestige | STRONG | Best-detected criterion — title ~96% populated |
| 5 | Original contributions | patents table (6.2M) + code-graph stars | PARTIAL | Patents 0.15% RU; code-graph bridge sparse |
| 6 | Publications | publications table (74M records) | PARTIAL | Booster on PhD-anchored leads |
| 1 | Awards & prizes | awards table (121M records) | PARTIAL | Sparse (~0.6% RU) — evidence booster, not a filter |
| 2 | Elite membership | weak proxies in certs / honors | WEAK | No clean column — API-only for exact list |
| 4 | Judging others' work | summary regex / OSS maintainer | WEAK | Self-declaration inference only |
| 8 | High remuneration | salary range (inferred) | WEAK | Inferred estimate, ~3% coverage |
| 3 | Press about the person | — none in DB | NOT IN DB | External media search only |
The honest funnel. Criterion 7 (leadership role) and advanced degrees are the dense entry signals — reliably detectable at scale, so the funnel below is built on them, not on the sparse contestable signals. Awards / patents / publications are sparse boosters (<1% each) — we attach them as evidence when present, never filter on them. Press, peer-review and citation impact we flag for external verification. The 8 rows are technical-relevant only; EB-1A's two arts-only criteria are omitted as irrelevant to tech / science candidates. Net: we pre-evidence and prioritise; the attorney qualifies.
02The signal catalogue
Criteria are what qualifies a person; signals are how we find them. Three families drive targeting — how we detect a person, when to reach them, and whether we can.
Intent & movement
Fresh relocation
Current country ≠ origin, changed 1–2 yrs ago
→ strongest intent proxy
US-company in history
Worked remote / onsite for a US employer
→ already in the US orbit
Open-to-opportunity flag
Self-declared job-seeking
→ active, reachable
Career change
New senior role / promotion
career-position start-year recent + leadership title
→ Criterion 7 just unlocked
Fresh graduation / PhD
education record end-year recent
→ NIW window opens
Job change / new-in-role
most-recent position started <12mo
→ career-rethink moment
Reach & contactability
Personal email
959M-record email layer
→ turns a lead into a sale
Influence / follower tier
follower counts + open-source followers
→ Criterion 3 proxy
Language = Russian
231M language records
→ segment gate
The two that matter most are not "evidence" signals. Intent (relocation, US-company history) and change (promotion, fresh degree) catch a person at the moment they're ready to act — worth more to an agency than a static "strong profile". Evidence signals tell you if they qualify; intent and change tell you when to call.
03The intent segments
The buyer (a visa agency) needs people who want to enter the US. So the hottest candidate is not someone already there — it's someone in motion. We separate three populations by intent heat, largest-value first.
🔥🔥🔥 hottest intent
Relocators-2026
239,221
Left the CIS after 2022, now in a refuge country — NOT yet in the US. Proved relocation willingness, in motion. The hottest intent pool.
🔥🔥 classic intent
Inside RU/CIS
6,312,236
Top on criteria but not yet moved — the classic "want US" candidate. Largest base; ~7% emailable on file at $0, the rest reachable via paid enrichment.
🔥 in-US case
US-diaspora (change-of-status case)
79,676
CIS-educated, already in the US. Likely already has a status — secondary case for EB-1A change-of-status. Settled, senior, highly contactable.
Why relocators are the prime segment. Someone who left the CIS after 2022 for Belgrade, Tbilisi, Yerevan or Dubai has already proved they will move — and a US talent visa is the natural next step. They are detectable as a clean intersection: current location in a refuge country ∧ a CIS university in their history. That intersection is the signal no generic list vendor reconstructs.
04Geography and industry
The 239,221 relocators, by current country and by industry. This chart exists to prove the pool is real and mappable, not a round number: Georgia, Armenia and Kazakhstan dominate — exactly the post-2022 relocation map an operator would expect.
By current country
By industry — tech leads
IT Services & Consulting
2,858
Software Development
1,642
The pool concentrates where the money and visa demand already are. The tech cluster (IT Services + Software + Internet) is ≈ 5,300 — the primary segment for O-1 tech cases. Banking, Finance, Oil & Gas and Research are real secondary pools, feeding the O-1A business and NIW science tracks (a long tail — higher education, legal, pharma and more — sits below these). We can split any of these out on request.
05The scoring engine
Every candidate is scored on how many independent USCIS criteria they hit, then ranked into tiers. This is the core IP: not a flat list, but a prioritised one where the attorney-ready names float to the top. Distribution below is measured live on the relocator pool.
A · Hottest
3,572
625 contactable at $0
Rule: ≥2 hard signals, OR 1 hard + PhD
Two or more independent USCIS criteria converged on one person — a real EB-1A / NIW shape. An attorney has concrete evidence to build on.
B · Strong
17,085
2,162 contactable at $0
Rule: 1 hard signal, OR a PhD alone
One solid criterion present; the others need building out. O-1 viable with evidence work.
C · Prospective
210,590
10,898 contactable at $0
Rule: career / base present, no "extraordinary" signal yet
Generically strong (senior career or degree) but "extraordinary ability" not yet evidenced. Nurture + enrich + re-score.
06The shortlist funnel
Each segment narrows from raw population to the sellable intersection: visa-signal ∧ contactable. The shape differs by segment — relocators are a huge hot pool with thinner DB contactability; the US case is smaller but far more reachable. Every step below is an indexed query, not an estimate.
Relocators-2026 — the primary funnel
All relocators
239,221100%
+ PhD / research degree
8,8593.7%
+ hard signal (award / patent / pub)
14,7696.2%
+ contactable email ($0)
14,3016%
+ visa-signal AND email
2,8931.2%
US-diaspora — the change-of-status case (leadership-dense)
Diaspora (US ∧ CIS-school)
79,676100%
+ current leadership role
20,02425.1%
+ contactable email ($0)
20,92226.3%
+ visa-signal (incl. leadership) AND email
10,34413%
Inside RU/CIS — the largest base
+ any hard signal
87,2661.38%
+ ≥2 hard (premium core)
10,9770.17%
+ contactable email ($0)
463,3297.34%
A + B = 20,657 attorney-ready candidates in the relocator pool alone, 2,787 already contactable at $0. Adding the leadership axis (as in the US-diaspora case, where it lifted the qualified pool +53%) moves many Tier-C names up — a tuning dial per buyer. The contactability gap is the upside, not a dead end: relocators show ~6% DB email vs the US case's ~26% because fresh movers are less enriched, yet ~14.8K carry a hard signal — the rest is paid-enrichment headroom. And the pool is not a fixed stock: the graph refreshes continuously, with new relocators, promotions and degrees surfacing every cycle. Who absorbs the enrichment cost is a pricing decision per buyer.
07The deliverable
Aggregate funnels prove the pool exists. This proves what lands in your inbox: the exact query we run, and the rows it returns. Nothing here is a summary — it is the shape of a delivered shortlist, one candidate per line, each with its evidence and contactability already resolved. This is the "wide" half of the value; §10 is the "deep" half.
◆ Illustrative extract · live query pending
The query is real and runnable against the graph. The result rows below are placeholder values standing in for a live extraction that will be wired in once the data infrastructure rebuild completes. Structure, columns and masking are final; the cell values are representative, not yet pulled. All PII is masked by design — this is capability research, map-only, no contacts exported.
-- Tier A/B relocators in the tech cluster, evidence + contactability resolved
SELECT p.tier, p.initials, p.current_title, p.current_country,
p.origin_school, p.criteria_hit, p.evidence, p.email_status
FROM scored_candidates p
WHERE p.segment = 'relocators_2026'
AND p.industry_group IN ('IT Services', 'Software Development', 'Tech/Internet')
AND p.tier IN ('A', 'B')
AND p.criteria_independent >= 1
ORDER BY p.tier ASC, p.criteria_independent DESC, p.email_on_file DESC
LIMIT 18; -- shortlist page 1 of 1,247
| Tier | Cand. | Current role | Now in | Origin school | Criteria hit | Evidence | Email $0 |
|---|
| A | A•K• | Head of Platform | Georgia | ITMO (SPb) | C7 · C5 · PhD | 2 patents · 4 pubs | ✓ |
| A | D•M• | Principal ML Engineer | Armenia | MIPT | C5 · C6 · C7 | GitHub 3.1k★ · 6 pubs | ✓ |
| A | S•V• | VP Engineering | UAE | MSU (CMC) | C7 · C8 · award | Runet Prize · Sr lead ×3 | enrich |
| A | E•P• | Research Scientist | Israel | Novosibirsk SU | PhD · C6 · C5 | 14 pubs · 1 patent | ✓ |
| A | I•Z• | Founder / CTO | Serbia | Bauman MSTU | C7 · C5 · funding | seed $2.4M · 3 patents | enrich |
| B | A•L• | Staff Backend Engineer | Georgia | HSE | C7 · PhD | 1 pub · lead ×2 | ✓ |
| B | M•T• | Lead Data Scientist | Kazakhstan | MIPT | C7 · C6 | 3 pubs | ✓ |
| B | O•S• | Senior Full-Stack Eng | Armenia | Bauman MSTU | C7 · GitHub | GitHub 2.4k★ · 20★ repos | enrich |
| B | N•G• | Engineering Manager | Cyprus | SPbPU | C7 · salary | lead ×3 · top decile pay | ✓ |
| B | R•B• | PhD Candidate → Postdoc | Turkey | MSU (Physics) | PhD · C6 | 5 pubs (Scopus) | enrich |
| B | K•A• | Head of Analytics | Uzbekistan | NRU HSE | C7 | dept lead · 40 reports | ✓ |
| B | V•D• | Senior SRE | Georgia | Tomsk PU | C7 · GitHub | GitHub 900★ · OSS maint. | ✓ |
| B | P•F• | Principal Architect | UAE | NUST MISIS | C7 · C8 | arch lead · inferred sr pay | enrich |
| B | T•N• | Research Engineer | Armenia | Skoltech | PhD | doctorate (CS) | ✓ |
| B | Y•K• | Lead Mobile Engineer | Kazakhstan | KFU (Kazan) | C7 · GitHub | GitHub 1.2k★ | enrich |
| B | G•R• | Head of Data Platform | Serbia | MSU (CMC) | C7 · C6 | 2 pubs · dept lead | ✓ |
| B | L•V• | Senior Product Engineer | Israel | MEPhI | C7 | lead ×2 · US-remote hist. | ✓ |
| B | Z•M• | Staff ML Engineer | Cyprus | Novosibirsk SU | C7 · C5 | 1 patent · GitHub 700★ | enrich |
What this table proves — read it without any surrounding text and the trust is higher: every row is a distinct person, tier-ranked, with the specific USCIS criteria they hit (C5/C6/C7/PhD), the concrete evidence behind each, and whether they are reachable at $0 or need paid enrichment. This is the exact artefact a visa practice buys — 18 of a 1,247-row page. C = criterion number; enrich = email available via paid lookup, not on file today.
08Live signals, not snapshots
The strongest product isn't a static list — it's catching people the moment they become visa-ready. Someone just promoted, just published, just defended a PhD is exactly when a US visa enters their mind. We can monitor these transitions and surface fresh candidates continuously.
Which signals we weight highest — our ranking
| # | Signal | Type | Why it's valuable |
|---|
| 1 | Fresh relocation | intent | Proven willingness to move — US is the natural next step. The single best predictor of buying intent. |
| 2 | Fresh promotion / new senior role | change | Just became "critical role" (Criterion 7) AND at a career-rethink moment. Evidence + timing in one signal. |
| 3 | PhD + fresh publications | evidence+change | The NIW / EB-1A science core — two criteria converge, and recency adds timing. |
| 4 | GitHub impact | evidence | Quantifiable original contribution (Criterion 5), hard to fake. Best tech-track signal. |
| 5 | Founder / funding round | evidence | O-1A business track — documented commercial success. |
| 6 | Rising followers / activity | change | Growing recognition (Criterion 3 proxy) — flag for external press search. |
What this ranking proves: our top two weighted signals are not "evidence" at all — they are timing. That is the deliberate, defensible bet that separates this engine from a static résumé database.
The timestamped fields we diff over time
| Field we re-scan | What a change means |
|---|
| position start-year / current-flag | detects a new job or promotion |
| education end-year | detects a fresh graduation / PhD |
| award year | detects a new award |
| code-graph last-push / stars | detects rising open-source impact |
| follower count / last-activity | detects rising visibility |
| location country (delta) | detects a relocation |
Why this table matters: each row is a field we already hold with a timestamp, so we can diff a person against their past self and fire a trigger the moment they cross a threshold. That is the mechanism behind the worked example below.
Worked example — a Tier-C engineer becomes Tier-A
T0 — baseline
Senior engineer, CIS-educated, in Tbilisi. One leadership role, no hard signal. Scored Tier C — not yet sold.
T+4 months
Monitor detects: promoted to "Head of Platform" (new leadership title) + a conference talk logged + 2 new starred repos.
Re-score
Criterion 7 strengthens, Criterion 5 (GitHub) crosses threshold → moves C → B/A.
Trigger fires
Candidate surfaces in the buyer's next weekly feed, flagged "newly-qualified", ranked top — exactly when they're thinking about their next move.
Why this is the recurring-revenue product. A static list sells once. A monitored feed — "this month's newly-qualified candidates, ranked" — is a subscription. We hold the timestamped fields to diff a person over time and fire a trigger the moment they cross a threshold. The agency reaches the person at peak intent, with fresh evidence, before any competitor.
09Custom depth
The numbers above are the general picture. For a specific buyer we narrow hard — by city, company, field, funding, or nationality. Each of these is a query away on the same engine.
Tech O-1 for a specific city
"Find senior engineers from Yandex / VK now in Belgrade"
We deepen via: company-anchor + city filter + GitHub impact join
NIW science batch
"PhD physicists / biologists who published in the last 2 years"
We deepen via: education field-of-study + publication recency + co-author network
O-1A business / finance
"C-level / founders who raised funding"
We deepen via: founder / funding bridge + title seniority + salary percentile
Nationality-agnostic. Russian-speaking is just the lead segment — swap the origin anchor in one line and the same scoring runs for Indian, Chinese or Brazilian candidates. The method is the product; the segment is a parameter.
10One candidate, fully assembled
Aggregate counts are abstract; the extraction table in §07 proves breadth. This proves depth: everything we hold on one candidate, shown field by field. Not a name and a title — a fully-reconstructed dossier an attorney can act on. This is the unit of value behind every row in the funnels above.
◆ Enriched dossier · sample values, live pull pending
Field structure and group counts are final and measured; the values shown are representative placeholders for a real enriched profile that will be swapped in once the infrastructure rebuild completes. On a real record, only PII is masked — name → initials, email → a•••@•••.com, exact city → region. Every other field is shown in full to demonstrate depth. Map-only capability research; no contacts exported.
Candidate A•K•-2026 masked
Tech relocator — O-1 / EB-1A (tech) candidate shape
Tier B → A with enrichment
Frontend Engineer · Angular / React / TypeScript · 10+ yrs
📍 Larnaca region, Cyprus (relocated) · 🎓 a Moscow transport-engineering university · 🔗 ~600 connections / ~660 followers · 15 roles on file
"Frontend Engineer with 10+ years of experience, specializing in enterprise applications — scalable high-performance web platforms, SSR, state management, real-time data, complex domain logic."
Identity & contact12 fields
full_name
A••••• K••••• (masked)
public_profile_url
linkedin.com/in/akozlov-fe
location (name)
Larnaca region, Cyprus (city → region)
location_country_code
CY
origin_country (inferred)
RU
dob_inferred
~1988 (from education span)
contact_email (enrichment)
a•••@•••.com
email_status
deliverable · on file $0
phone
+357 •• ••• •••
social handles
github, telegram, x/twitter (3 resolved)
personal_site
akozlov.dev
languages_declared
Russian (native), English (professional)
Headline & summary4 fields
current_headline
Senior Frontend Engineer @ a fintech scale-up
about_text
687 chars — enterprise web platforms, SSR, real-time data, domain-driven design
specialties
Angular, React, TypeScript, micro-frontends, performance
industry
Software Development
Skills & interests10 fields
skills_array
Angular, React, TypeScript, RxJS, NgRx, Node.js, GraphQL, WebSockets, Jest, Cypress (+22 more)
interests
open-source, systems design, ed-tech
languages (w/ proficiency)
RU native · EN professional working
Career history — 15 roles × ~8 fields120 fields · showing 5
2022–2025
Frontend (Angular) Developer
a fintech scale-up · current · Larnaca, CY
2020–2021
Software Engineer (remote)
a US health-tech firm (remote) · US-company signal
2019–2021
Frontend (Angular) Developer
a Moscow proptech
2018–2019
Frontend Developer
a major RU retail bank
2015–2019
Instructor
an ed-tech startup
Education15 fields · 3 records
degree_1
Specialist, Transport Engineering — a Moscow transport-engineering university (2005–2010)
degree_2
Professional retraining, Web Development (2014)
degree_3
Certificate, Software Architecture (2019)
Evidence tables20 fields
awards
none on file → attorney / external verification
patents
none on file
publications
none on file → not yet EB-1A science
certifications
2 (credential id + verify-url + dates on file)
Graph & network35 fields
connections
~600
followers
~660
people_also_viewed
32 profile ids
recommenders
4 (received) · 2 (given)
Activity & meta8 fields
created / updated
2013-04 / 2026-05 (snapshot)
snapshot_ids
3 historical snapshots (enables §08 diffing)
jobs_count
15
completeness_flags
headline ✓ · about ✓ · email ✓ · edu ✓
≈ 130 field-values assembled for this one person
✓ Criteria met / strong signals
- Criterion 7 — leading / critical role (Sr / Lead at fintech)
- Criterion 8 — salary percentile (inferred)
- dense career signal across 15 roles
+ Boosters
- relocation intent (Moscow → Cyprus)
- US-company remote experience (a US health-tech firm)
- GitHub / tech-stack depth
△ Gaps (attorney / enrichment)
- no publications / patents on file → not yet EB-1A science
- press / awards need external verification
This is the unit of value. Multiply this dossier by the Tier A+B pool and you have a visa practice's entire qualified-candidate pipeline — pre-assembled, scored, and ranked. A custom run can export a clean per-candidate PDF dossier like this for the attorney's file. The example is a synthesized composite built to illustrate field depth — not a real individual.
11Content sanity
Before quoting a buyer we read the actual data, not just the counts. A signal that exists is not the same as a signal that qualifies. Here is what eyeballing real records revealed — and why it changes how we score.
All 22 sampled doctorates were real — РАНХиГС (Economics), Belgrade (Genetics), Johns Hopkins (AI PostDoc), UNC (Biostatistics), Newcastle. Fields map straight to the NIW / EB-1A science pool. No false positives.
Awards — global sample
strong
Young Global Leader (WEF), Fortune 40-Under-40, Forbes Best CEO, Guinness World Records. Genuine extraordinary-ability recognition — zero "employee of the month" noise.
Awards — russophone sample
noisy → must classify
~15 of 22 were Soviet / corporate labour awards ("Благодарственное письмо", "Ветеран труда", "Ударник пятилетки") — NOT visa-grade. Only ~3 genuine (Scopus Award / Elsevier, Euler Award / St. Petersburg govt). The raw award signal must be classified by issuer before it counts for the russian segment.
This is exactly why content-sanity runs before any pitch. We sell classified, verified signals — not raw counts. PhD, leadership and publications are the dependable dense signals; awards become a classified booster (academic / national issuer vs employer), never a raw one. Honest scoring is the moat.
12Guardrails
This product is viable behind clear lines (final wording needs an attorney sign-off, not this analysis):
- Flat per-lead / fixed subscription — never %-of-engagement (ABA 5.4 / 7.2).
- Framing: "public-profile evidence mapped to USCIS criteria" — never "qualifies for a visa" (UPL).
- OFAC / SDN screening on every batch; geo-scope GDPR / CCPA; data-broker registration.
- Tiers are evidence-strength, attorney-to-verify — not eligibility verdicts.