Ілля

Illia Naumenko
SYSTEMS ARCHITECT · AI & HIGH-PERFORMANCE BACKEND ENGINEER

Köln, Germany · [відкрити контакти](див. вище в блоці «контактна інформація») · [відкрити контакти](див. вище в блоці «контактна інформація») · [відкрити контакти](див. вище в блоці «контактна інформація»)

SUMMARY

Independent Systems Architect specialising in institutional-grade, AI-driven platforms for FinTech and GovTech domains. Over the past year,
independently designed and delivered two complex systems end-to-end: an algorithmic energy trading & risk protocol (AMVA-TRP) for CEE
energy markets and an AI anti-corruption analytics engine (ANTIA-CORP) for government procurement monitoring. Core expertise spans
multi-tier polyglot architectures (Go · Python · Node.js), vector-based predictive analytics (HNSW / Cosine Similarity), autonomous LLM/RAG
agents, and real-time data pipelines. Focused on domains where standard tooling fails — building proprietary mathematical cores and AI
agents that compete with institutional-class commercial software.

EXPERIENCE

Independent Systems Architect Jul 2025 – Present
Self-employed · Remote

Owned full delivery of two institutional-grade platforms as sole architect and lead engineer — from requirements analysis and system design
through polyglot implementation (Go / Python / Node.js) to Kubernetes deployment.
Designed 4-tier hybrid architectures: low-latency Node.js execution layer (<50ms), Python analytical core (Monte Carlo simulation, EWMA
covariance, Markowitz portfolio optimisation), autonomous LLM/RAG agents, and a Go mathematical engine with custom HNSW vector search
across 3.6M+ historical data points.
Engineered domain-specific mathematical frameworks: V-State multi-variable market analysis (cosine similarity), Mean-Variance portfolio
optimisation, and 512-dimensional tensor-based corruption risk scoring with precision ≥92% on verified test datasets.
Built real-time risk management systems with proactive 3–7 day forecasting, 24/7 AI-powered news sentiment monitoring, automated cross-border
arbitrage detection, and emergency circuit-breaker controls with 2FA approval flows.
Architected a three-level anti-corruption pipeline integrating OCR document parsing, Graph Network Analysis (Neo4j) for hidden affiliation
detection, and Explainable AI verdicts legally admissible as primary evidence for official investigations.

Software Developer — Volunteer May 2024 – Jul 2025
Donater · Mykolayiv, Ukraine / Remote

Built the backend architecture for a charitable Telegram bot supporting internally displaced persons, leveraging NestJS and OpenAI API to deliver
AI-driven assistance at scale.
Engineered efficient storage solutions with MariaDB and Redis, improving system reliability and high availability for end users.
Contributed to an Agile development lifecycle, maintaining version control standards and participating in cross-functional code reviews.

E D U C AT I O N

B.Sc. · Computer Software Engineering B.Sc. · Operation of Ship Electrical Equipment
Admiral Makarov National University of Shipbuilding Admiral Makarov National University of Shipbuilding
Aug 2025 – Present · Enrolled in Year 2 · Distance Aug 2023 – Present · Mykolayiv, Ukraine / Remote

LANGUAGES

English German Ukrainian Russian
Advanced (C1) Intermediate (B1) Native Native

SKILLS

Languages & Runtimes: AI / ML / Vector:

Go (Golang) Python Node.js TypeScript JavaScript RAG Pipelines LangChain HNSW Milvus Pinecone

SQL Bash OpenAI API Gemini API Anthropic LlamaIndex MLOps

Backend & Data: Scientific Computing:

FastAPI NestJS Microservices gRPC RabbitMQ Gonum NumPy SciPy Pandas PyTorch Qiskit

PostgreSQL TimescaleDB Neo4j Redis TypeORM PennyLane

Prisma

DevOps & Cloud: Frontend:

Docker Kubernetes Terraform AWS GCP React React Native Electron Vue.js D3.js WebSockets

GitHub Actions Prometheus Grafana Expo
PROJECTS

AMVA-TRP — Autonomous Multi-Variable Algorithmic Trading & Risk Protocol FINTECH · ETRM

Go · Python (FastAPI, NumPy, SciPy, Backtrader) · Node.js (NestJS) · Electron · React Native · TimescaleDB · Redis · Kubernetes
4-Tier Hybrid Architecture for CEE Energy Markets: Institutional-grade algorithmic trading system for gas & electricity markets (HUDEX, TTF,
VTP, HUPX, EEX). Low-latency Node.js microservice layer (<50ms order execution, 50K ticks/sec), Python analytical brain (Monte Carlo,
EWMA covariance, Markowitz portfolio optimisation), autonomous LLM/RAG agents, and a Go math engine 10–11× faster than NumPy
equivalents.
V-State Predictive Analytics: Proprietary multi-dimensional Market State Vector framework encoding price, weather, FX, and implied volatility.
Custom HNSW cosine similarity search across 3.6M+ historical snapshots identifies analogous market configurations in <5ms, generating
probabilistic forecasts and arbitrage signals with confidence scores.
Autonomous AI Market Intelligence: 24/7 LLM agents (RAG + fine-tuned NLP, sentiment −5 to +5) monitoring Bloomberg/Reuters feeds and
regulatory portals, automatically adjusting VaR coefficients and generating real-time Netback profit calculations across 5 European energy hubs.
Proactive Risk & Liquidity Management: Monte Carlo ±3σ/±5σ stress-test modules forecast margin calls and cash flow gaps 3–7 days ahead.
Mobile executive dashboard with Emergency Stop and 2FA hedge approval with <500ms strategy switching.

ANTIA-CORP v2.1.0 — AI-Powered Anti-Corruption Analytics Platform GOVTECH · PREDICTIVE ANALYTICS

Go · Python (FastAPI, PyTorch, spaCy, Tesseract OCR) · Node.js · PostgreSQL · Milvus · Neo4j · RabbitMQ · Kubernetes
Three-Level Analytical Pipeline (L1→L2→L3): High-throughput GovTech platform for real-time monitoring of Ukrainian government
procurement (Prozorro), processing 10,000+ tenders/day. Go tensor filtering (MV-Core, <50ms) → Python NLP + OCR + semantic compression
(NLP-S, <5s) → LLM/RAG verdict with XAI (CAI-A, <15s). Precision ≥92% on verified datasets.
512-Dimensional Corruption R-Score: Proprietary tensor risk scoring across four domains: price anomalies, temporal patterns, semantic
fingerprints in procurement documentation, and historical reputation profiles. HNSW ANN search against a 10-year archive of verified corruption
cases.
Graph Network Analysis (GNA): Neo4j-based engine exposing hidden affiliation networks via Community Detection (Louvain), Shortest Path,
PageRank, and Temporal Analysis — identifying shell companies, ownership chains, and coordinated bidding rings across procurement
databases and UBO registries.
Explainable AI & Legal Compliance: Mandatory Chain-of-Thought XAI module with full factor decomposition per verdict, making outputs
admissible as primary evidence for official investigations. GDPR-compliant. Dual-Storage Protocol (PostgreSQL + Milvus), K8s HPA, RPO ≤1h /
RTO ≤4h.

TenderMind AI — Intelligent Global Tender Analysis Platform B2B SAAS

NestJS · TypeScript · Python · PostgreSQL · OpenAI API · LangChain · Pinecone · React

Hybrid Search Engine: SQL filtering (budget, region, CPV codes) combined with semantic vector search. “Synthetic Tender” technique
transforms user interviews via GPT-4 into professional specs, maximising retrieval precision across Prozorro, TED, GOV.UK, and Doffin.
AI Automation Suite: Autonomous onboarding agent for structured user profiling, RAG pipelines for legal document generation and risk
assessment, real-time server-side profitability & logistics calculations. Geospatial market maps and competitor analytics dashboard.

AI Legal Assistant — Real-Time Judicial Registry Access LEGALTECH

Node.js · Puppeteer · OpenAI API · PostgreSQL · Telegram Bot API

Puppeteer-based RPA layer overcoming absent public APIs in Ukrainian judicial registries — handling session emulation, anti-bot evasion, and
form navigation. Semantic analysis on scraped data generates precedent summaries and strategic recommendations via a conversational
Telegram interface (e.g., “Find fraud precedents from 2023”).

Digital Ecosystem for Maritime Recruitment HR PLATFORM

NestJS · TypeScript · PostgreSQL · Redis · OpenAI API · React · Telegram Bot API

Dual-sided marketplace connecting seafarers and recruiters with AI-driven candidate evaluation (gap analysis, vacancy matching), RBAC
security, live chat, and a natural language Telegram bot for recruiter queries. LLM integration for applicant screening and a personalised AI
coach improving hire rates.

C E R T I F I C AT I O N S & C O U R S E S

IBM Fundamentals of Quantum Algorithms / Coursera IBM AI Developer / Coursera
Issued Jan 2026 · ID: 87c569df-0b10-46f3-869e-ab22e24f187f Issued Apr 2025 · ID: JB4ERI6SVSMP

Amazon Junior Software Developer / Coursera IBM DevOps and Software Engineering / IBM Quantum
Issued Apr 2025 · ID: BL5KAOOBEOC8 Issued Apr 2025