An Experimental Study of Edge Deployment Strategies for Cryptocurrency Analytics Workloads
ComputingSpringer Nature
Full-Stack Software Engineer
I design and ship production SaaS, evidence-backed AI agents, and on-chain systems with verifiable decisions, payments, and outcomes.
01 / Publications
Peer-reviewed work spanning edge computing and cryptocurrency analytics.
ComputingSpringer Nature
02 / AI Agents & Web3
My 2026 focus: evidence-backed agents + x402 agentic payments + verifiable on-chain outcomes. Ten-plus hackathons and ecosystems across Arc, Base, BNB, Casper, Creditcoin, Somnia, and Starknet.
1st Place · $10k grand prize
Canteen × Circle × Arc · Top 1 team · project keryx
Standouts Winner · $50k prize pool
Canteen × Circle × Arc · Top 10-12 teams · project reasoning-receipt
3rd Place · Strategy Skills · $6k pool
CoinMarketCap × Trust Wallet · 21 teams · $36k total · project undertow
Agora Agents Hackathon · Canteen × Circle × Arc
Standouts winner. An AI oracle paid over x402 for prediction markets, where every price ships with a Merkle-rooted reasoning DAG (byte-verifiable, SHA-256 per node), settled on Arc in <1s (~$0.0007/receipt). Ensemble of 5 agents: Bull/Bear/Edge (parallel Gemini, isolated context) feed a Supervisor (weighted-Bayesian + Brier priors), then a Critic. 4,558 on-chain receipts · 217 markets · cross-chain CCTP V2.
Lepton Agents Hackathon · Canteen × Circle × Arc
1st place at the Lepton Agents Hackathon. A live citation-toll research agent that buys exact article versions over x402, rewards cited creators in USDC, and exports portable receipts for its evidence, decisions, and settlement. v0.17 adds claim-aware source portfolios and citation-freshness tracking.
Orion Builder Hackathon · Base Sepolia
Evidence layer for agent marketplaces. It seals claim-specific adversarial trials before execution, scores observations with deterministic assertions, and produces signed, tamper-evident receipts with independently verifiable EAS anchors.
DreamDEX · Somnia Shannon Testnet
Policy-enforced BTC/ETH portfolio protection using short-duration Event Contracts. It preserves linked intent-to-execution receipts and has public testnet evidence for fill, settlement, winning redemption, and position reconciliation.
Casper Agentic Buildathon 2026
Reputation-staked x402 signal oracle + DeFi agent on Casper.
BNB Hack: AI Trading Agent Edition · CMC × Trust Wallet
3rd place, Strategy Skills track. An AI-agent trading skill exploiting the gap between crowd sentiment and positioning. Backtested and reproducible, native to the CMC Agent Hub (MCP · x402 · CLI · Skills).
On-device private alpha agent for crypto traders. A local LLM + RAG via @qvac/sdk turns watchlists and positions into signal reports. Zero cloud, $0 inference, fully offline.
Published to npm
AI-agent skills that post and engage across 12 social platforms (Twitter, LinkedIn, Reddit, Medium, Bluesky, HackerNews, ProductHunt…). Built for Claude Code / Gemini CLI with multi-thread posting workflows.
03 / Selected Work
2026
Maintainer-governed admission and evidence layer for AI-assisted open-source contributions. Read-only by default; maintainer consent, scoped permits, deterministic evidence, validation, and human review gate every draft proposal. The maintained Rust/Tokio core supports multiple LLM providers, MCP, and an authenticated dashboard, with 686 passing tests.
2025 - Present
Multilingual blog & SEO SaaS. Full system architecture with a Node.js + Python backend, PostgreSQL, REST API, a React + TypeScript frontend, an SEO engine with auto-crawl, and a Dockerized deployment on Linux.
2024 - 2026
Multilingual (VI + EN) AI reading platform on Astro 5 SSR + React 19 islands. Better Auth, bookmarks, four themes, full-text search, and full SEO (sitemap + JSON-LD), edge-first on Cloudflare D1 / R2.
2026
Self-hosted AI "personality clone" that learns from your chat history to mirror your voice. BYOK multi-LLM (Claude/GPT/DeepSeek/Ollama), runs 100% locally (privacy-first), integrates Discord/Telegram/WhatsApp. Open-source (MIT).
2025
Real-time WebSocket signals (Binance/Bybit), multi-account execution, PnL dashboard, and risk management, Dockerized with 24/7 auto-restart.
2026
Reverse-engineered guide to X/Twitter's recommendation algorithm with practical playbooks for maximizing reach.
2025
Turns scanned PDFs into searchable text entirely in the browser: no upload, 100% client-side with Tesseract.js.
2025
Native desktop app for disk management: format, partition, extend, and shrink USB / SD / HDD volumes.
2026
Native desktop app that auto-runs your project GitBash sessions on machine startup.
04 / Experience
2025 - Present
VentureVibePressNet
Multilingual blog & SEO SaaS platform. Owned system architecture, Node.js + Python backend, PostgreSQL, and REST API; React + TypeScript frontend; SEO engine with auto-crawl; Dockerized deployment on Linux.
2025
VentureFastBuff (Digital Marketing SaaS)
Built an automated order-processing platform with multi-platform API and payment integration, generating steady revenue from business customers on a Node.js + React stack.
2023 - Present
IndependentPersonal product ecosystem
Sole sysadmin for 10+ production systems running 24/7 at near-100% uptime. Linux, Docker/Compose, DNS zones, mail (MX/SPF/DKIM/DMARC), DB backup & recovery (PostgreSQL/MongoDB), SSL/TLS, Nginx reverse proxy, firewall, and monitoring.
05 / Open Source
Pull requests I've written and merged into open-source projects. Every PR is real, reviewed, and landed upstream.
| Repository | Stars | Merged |
|---|---|---|
| Osmantic/ODS ↗Local AI server stack | 6.1k★ | 66 |
| lutris/lutris ↗Open-source game manager for Linux | 10k★ | 1 |
| amanusk/s-tui ↗Terminal CPU stress and monitoring UI | 5.1k★ | 1 |
| koala73/worldmonitor ↗Global intelligence dashboard | 85k★ | 4 |
| soxoj/maigret ↗OSINT username search across 3000+ sites | 37k★ | 3 |
| Orchestra-Research/AI-Research-SKILLs ↗Open-source AI research skills library | 12k★ | 1 |
| falconry/falcon ↗Python web API and microservices framework | 9.8k★ | 1 |
| HolmesGPT/holmesgpt ↗CNCF Sandbox AIOps investigation agent | 3.2k★ | 1 |
06 / About
I'm a full-stack engineer with 4+ years building and shipping production software end-to-end: SaaS platforms, developer tools, and, most recently, autonomous AI agents. I learn fast, work independently, and care about the whole path: from the first line of code to a system that stays up in production.
Author of ContribAI (244★, 686 tests, Rust/Tokio) and creator of Keryx, 1st-place winner at the Lepton Agents Hackathon 2026. Across GitHub: 112 public repositories and 97 reviewed pull requests merged into upstream projects.
From API design and Node.js/Python/Rust backends to React & Astro frontends. I take SaaS products from zero to production, then keep them running.
Evidence-backed agents with explicit policy, portable receipts, and verifiable on-chain payments. That thread runs through ContribAI, Keryx, AgentTrial, and my 2026 hackathon work.
Sole sysadmin for 10+ production systems at near-100% uptime: Linux, Docker, DNS, mail, SSL/TLS, backups, and monitoring, all self-managed.
07 / Toolbox
Seven-plus languages across the stack, picked per problem, not per résumé.
08 / Education & Research
B.Sc. in Computer Science
School of Information and Communications Technology
Graduated · 2021 - 2025Coursework projects (public repos): a web novel reader, a training-management system (QLDT), and 10+ Android/Kotlin apps.
Replication data + analysis scripts for the paper "Large Language Models for Software Engineering: A Bibliometric Mapping Study".
Research benchmarking AI-agent security alongside trading-strategy analysis.
Languages
Vietnamese: Native · English: Working proficiency (technical docs, international clients & hackathons)
09 / Contact
Open to work and interesting collaborations in full-stack, AI agents, or Web3. The fastest way to reach me is email.