Hakim Tohirovich
AI SCIENTIST-ENGINEER · INDEPENDENT RESEARCHER
Independent researcher with 20 years of cross-domain work across 7 scientific disciplines. Founder of ONTO Standards Council. Built the epistemic discipline standard for AI — from research to production.
📞 Available for video call · In-person meeting
Background
20 years of independent cross-domain research — mathematics, epistemology, medicine, law, physics, cybersecurity, biology — produced a knowledge system that became the foundation for ONTO: the epistemic discipline standard for AI. Every component — research, engineering, deployment — built by one person.
Output: 12 published research reports · open-source scoring engine (993 lines, reproducible) · 22+ models evaluated across 5 domains · cited in government tender documents for 9 countries.
EDUCATION
Self-directed research program
No formal AI/ML programs existed in Central Asia when this research began (2005). 20 years of independent study across mathematics, epistemology, medicine, law, physics, cybersecurity, biology. Equivalent output: 12 published reports, open-source scoring engine, peer-verifiable methodology adopted in government tender documentation across 9 countries.
TECH STACK
LANGUAGES
PythonRustJavaScriptHTML/CSSSQLBash
AI / ML
LLM ArchitectureSystem Prompt DesignRAGInference OptimizationModel EvaluationScoring SystemsEpistemic Scoring
BACKEND & INFRASTRUCTURE
FastAPIPyO3REST APISSEStripeRailwayGitHub PagesCI/CDNginxDocker
SECURITY & CRYPTO
Ed25519Merkle TreesForensic WatermarkingProof ChainsJWT
RESEARCH DOMAINS
EpistemologyMathematicsMedicineLawPhysicsCybersecurityBiology
TIMELINE
2005–2015
Research Foundations
Self-directed cross-domain research. Mathematics, epistemology, philosophy of science. No AI/ML programs existed in Central Asia.
2015–2023
Formalization
R1-R7 epistemic discipline framework. Cross-verified across 7 scientific domains. Corpus construction: 169 files, ~900K tokens.
2020–2023
Discovery
Loaded corpus into AI — models changed. Started citing sources, saying "I don't know." Without modification. GOLD Core was born.
2023–2024
Engineering
Scoring Engine v3 (Python). onto_core (Rust). FastAPI backend. API, proxy, auth, billing. Ed25519 proof chain.
2024–2025
Validation & Launch
22+ models evaluated. 12 reports published. Agent v5. Dashboard. 5 production websites. R8-R16 cognitive architecture (Human AI protocol). IT Park residency.
2026
Production & Global Outreach
ONTO production-ready. Government tenders to 9 countries. Venture applications. AI Alliance UZ member.
PROFESSIONAL ROLES
AI/ML Research Scientist
R1-R16 · 169 files · 7 domains · 12 reports
20y
AI/ML Engineer
Scoring v3 · 993 lines · 22+ models · EM1-EM5
3y
AI Security & Cryptography
Rust · Ed25519 · Merkle · 104-byte forensic
2y
Backend & Infrastructure
FastAPI · ~5500 lines · REST · SSE · Stripe · Railway · CI/CD
3y
Data Engineer
900K token corpus · scoring pipelines
3y
AI Product Engineer
6 roles via AI · 5 prod sites · 9 countries
2y
RESEARCH PROJECT — SHIPPED TO PRODUCTION
Built end-to-end: research → engineering → deployment. Both products run on GOLD Core — 169 files, 900K tokens, 20 years of research. One integration covers both.
FOR REGULATORS
PRODUCT 1
Regulator API + Dashboard
ONTO Standard — AI quality grading for any country
DASHBOARD
Every AI graded A-F
All AI models on one screen. Trends. Domains. Violations.
ENFORCEMENT
Cryptographic proof chain
Ed25519 signed. Tamper-proof evidence for regulators.
A-F
GRADES
104B
PROOF SIZE
$0
PILOT COST
✅ 100% PRODUCTION READY
Regulator Dashboard →
FOR AI PROVIDERS
PRODUCT 2
Human AI API
Cognitive architecture for any AI model
R1-R7
Epistemic discipline
Cites sources. Quantifies confidence. Says "I don't know."
R8-R16
Cognitive architecture
Disciplined creativity. Causal reasoning. Self-awareness.
16
RULES
10×
IMPROVEMENT
$0
MODIFICATION
✅ PROTOCOL COMPLETE · PLATFORM IN PROGRESS
LIVE PROOF
ONTO Agent
Compare any AI with and without discipline. 5 languages. 30 seconds. No signup.
Try it →
HOW IT WORKS
Like a currency converter. Raw input in — disciplined output out. Same model. One layer.
FORMULA
Any AI
+
GOLD Core
=
Disciplined AI
SCHEMA
Request
any question
→
GOLD Core
conversion layer
→
Disciplined answer
sources, confidence, proof
CIRCLE
Any AI
↓
GOLD Core
conversion
↓
Quality
×10
↓
Compliance
A-F grade
↓
Sectors
7 domains
One idea — three ways to explain it.
Formula — for a slide, a business card, one sentence.
Schema — for an engineer. Input → process → output.
Circle — for a minister. One investment, three results.
Schema — for an engineer. Input → process → output.
Circle — for a minister. One investment, three results.
CONVERTER IN ACTION — REAL API OUTPUT
WITHOUT GOLD
"Studies show that vitamin D is very effective for preventing respiratory infections. Many experts recommend supplementation."
⚠ No source · No confidence · No limitations · No DOI
WITH GOLD
"Vitamin D supplementation shows moderate effect (OR 1.3, 95% CI 1.1-1.5) on acute respiratory infections. Source: Martineau et al., BMJ, 2017. DOI: 10.1136/bmj.i6583. Grade: systematic review, 25 RCTs. Confidence: moderate. Limitation: heterogeneity across study populations."
✅ Source · DOI · CI · Grade · Limitations · Confidence
Same model. Same question. One layer difference.
PUBLISHED RESULTS
169
files · GOLD Core
900K
tokens · 7 domains
993
lines · scoring engine
12
published reports
22+
models evaluated
5
production websites
9
countries in outreach
104B
Ed25519 proof chain
Composite Score
7.1×
0.12
8.85
Sources cited
0→4
0
4
Confidence disclosed
0→56%
0
56%
Unknowns disclosed
0→4
0
4
Unknown recognition — the edge
26×
0.04
0.96
■ WITHOUT DISCIPLINE ■ WITH ONTO GOLD | Zero retraining. Inference-time only.
PUBLICATIONS & OPEN SOURCE
CASE STUDY × 4
CS-2026-001 — CS-2026-004: AI Epistemic Quality Evaluation
FIELD OBSERVATION × 2
FO-2026-001, FO-2026-002: Model Behavior Under GOLD Discipline
ARTICLE
From Golden Ratio to GOLD Core: The Mathematics of AI Discipline
OPEN SOURCE
Scoring Engine v3 — 993 lines, deterministic, reproducible
WHITEPAPER
WP-2026-002: Architecture, Economics, Risk Analysis
LANGUAGES
Russian
NATIVE
English
WORKING · READ / WRITE · SPOKEN IN PROGRESS
EXPERTISE
Research
AI safety, epistemic quality, model evaluation, cross-domain methodology
Engineering
AI/ML systems, scoring pipelines, proof chains, production APIs
Advisory
AI governance, regulatory standards, epistemic discipline, quality frameworks
Consulting
AI quality audits, model benchmarking, compliance grading, GOLD integration
Tashkent · Remote
📞 Open for conversation · Video call · In-person meeting
ONTO infrastructure: API, scoring engine, proof chain — fully automated. Runs independently.
English: Working proficiency — read, write, technical documentation, research papers. Speaking — in active development. Every document on this page was written in English by the author.