🔗 n30sk.cc — systems that check claims

📫 [email protected]

LinkedIn

CV

Senior Frontend Engineer with 10+ years across fintech, crypto trading, and enterprise — including team lead experience and multiple 0→1 product builds.

I specialize in React, Next.js, and TypeScript, with focus on scalable frontend architecture, user-critical flows (authentication, KYC, onboarding, account management), and performance-conscious delivery. Comfortable across the stack: Node.js, REST/GraphQL, data-layer design, and mobile (React Native).

• Production-grade frontend architecture (Feature-Sliced Design, App Router, SSR)
• Security-critical user flows — auth, KYC, 2FA, API key management
• Scalable state and data-layer design with TanStack Query
• Testing strategy and regression safety across critical flows
• Cross-functional delivery in fintech and trading domains

Frontend
React · Next.js · TypeScript · JavaScript · GraphQL · Tailwind · CSS Modules · styled-components · Zod

State & Data
TanStack Query · Redux Toolkit · MobX · Jotai · Recoil

Backend & APIs
Node.js · NestJS · Express · PostgreSQL · MongoDB · REST APIs

Testing
Jest · Testing Library · Vitest · Cypress

Architecture
Feature-Sliced Design (FSD) · Component-driven architecture · SSR / App Router · performance optimization · accessibility (a11y)

Tools
Vite · Webpack · Docker · OpenShift · GitOps · CI/CD · Agile/Scrum

Mobile
React Native · Cordova

• Clean architecture that stays healthy as products scale
• Business-impactful frontend work — conversion, reliability, performance
• Honest engineering: maintainable code over impressive-looking complexity
• Reliable testing as a first-class practice, not an afterthought
• Cross-functional collaboration with backend, product, and design

My pet project — Biassemble 🧠 An AI that finds cognitive biases in your story — and must prove every finding with a verbatim quote from your own text. Found nothing? It says nothing. Silence is a valid result, not a failure.

⚙️ The flow: [FE · React] → [BE · Fastify] → [Core · TypeScript] ⇄ [Engine · Python] story → hybrid retrieval (LLM ∪ vector search on #pgvector) → assessment (#Gemini as the LLM) → findings + exact quotes

🔗 Live: https://frontend-topaz-eight-10.vercel.app

📦 Code:

biassemble — the app: frontend + API

biassemble-core — the brain: prompts, assessment logic, evals

biassemble-engine — retrieval: 38-bias knowledge base, embeddings, search

Built with real eval discipline: golden datasets, CI quality gates (false-positive rate < 0.10), versioned prompts, Recall@5 tracked openly — including the gap still being closed.

#AI #LLM #RAG #TypeScript #Python #SideProject

🔗 Async URL Status Checker — AI-powered
NestJS + React + TypeScript. Submit a list of URLs; the backend checks each via background HTTP HEAD requests (concurrency-capped, cancellable) while the frontend tracks live progress.

📦 Code: 3205

🔗 Bingo — test assignment
Recent technical assignment — good signal for code style and engineering approach.

The Model's Explanation Had the Right Answer. Its Verdict Didn't.

Building a RAG Engine: Three Engineering Assumptions I Had to Unlearn

My Bias Detector Found "Cherry-Picking" in the Answer "No Info"

🔗 All articles on dev.to