I build compliance infrastructure and the products around it, for teams that can’t afford to be wrong about who they’re doing business with.
SadatNazarli
About Sadat Nazarli
Every payment, every onboarding, every new partner starts with one question: can we trust them? I build the systems that answer it. Verifex screens people and companies against sanctions and PEP sources, across alphabets, and returns a decision you can defend: what was checked, what wasn’t, and why. Summativ brings that same rigour to how schools in Azerbaijan assess their students.
Verifex in numbers
47
Sanctions, PEP and registry sources catalogued
97.3%
Exact-name recall on real listed parties, with the confidence interval published
0 / 5,000
Adversarial near-miss names that raised a severe false positive
300 / 300
Listed sanctions parties in the tested sample surfaced as actionable
Verifex, self-administered benchmark measured 12 Aug 2026, published on verifex.dev/benchmark
Ventures & work
Three chapters. One obsession: getting it right.
Keep scrolling →
01 Founder · Jan 2026 — Present
Verifex
A sanctions & PEP screening API that gives compliance teams answers they can defend.
47 sources catalogued, from OFAC, UN, EU and UK lists to GLEIF and UK Companies House
Fuzzy, phonetic & transliteration matching across alphabets
Ownership chains screened, with OFAC 50% Rule calculations as analytical support
An evidence record for every decision: which sources were checked, and which weren’t
A missing source withholds clearance, so incomplete work never reads as clear
Assessment infrastructure for Azerbaijani educators.
Small and large summative exams (KSQ / BSQ), exam access, student data, grading and result analytics in one system, so a school’s assessment stops living in scattered spreadsheets and group chats.
Wrote the API documentation other developers build on
Automated manual data entry with Telegram bots
01Application
02KYC checks
03Compliance APIs
04CRM record
05Merchant live
Try the engine
Same person. Different alphabets.
The hard part of screening isn’t exact matches. It’s knowing that Садат Назарли, Sədat Nəzərli and Sedat Nazarlı are one person, and that Sadat Nazarov probably isn’t. This is a pocket-sized version of that logic. Type any spelling.
Name screening — liveref: Sadat Nazarli
100/100
Match
01 Script
Cyrillic
02 Normalized
sadat nazarli
03 Tokens
2/2 aligned
04 Folding
Transliterated
A toy version of the matching problem Verifex solves in production: one person, many alphabets, and every decision explained.
How I build
Principles
01
Recall comes first.
In sanctions screening a missed match is a regulatory breach, not a bug. Verifex tunes for recall first and publishes its benchmark with the limits attached: what was measured, on what, and what it doesn’t prove.
02
Every decision leaves evidence.
Each check records which sources it ran against, which list versions, and which sources it couldn’t reach, so any answer can be reproduced in front of an auditor.
03
Trust shouldn’t need an enterprise contract.
Self-serve from $49 a month, with free screens to start. A two-person fintech deserves the same protection as a bank.