I am a self-taught, product-minded developer transitioning from a strong business background (BTEC Business, 2024) to Software Engineering (B.Sc. in IT at UIT VNUHCM, 2026–2028).
Instead of just writing code, I focus on solving real user pain points and creating leverage through automated systems. My background in business analysis allows me to bridge the gap between complex codebases and commercial/user value—ensuring every feature built serves a direct product objective.
📍 Ho Chi Minh City, Vietnam
🎓 UIT – VNUHCM (B.Sc. IT, Distance Learning - Strategic choice to balance academic growth with active development)
💼 Focus: Product-Minded Software Engineering · AI Integrations & Developer Tooling · Quantitative Trading Systems
🏆 Objective: Building high-utility automated products that streamline complex industry workflows
Here are featured projects that highlight my product mindset, technical versatility, and ability to build complete end-to-end solutions:
🦜 VietFi Advisor — AI Financial Assistant for Gen Z (WDA 2026)
- The User Pain Point: Personal finance apps in Vietnam lack local market context (gold rates, local banking interest, VN-Index metrics) and fail to address the credit card debt/margin trap that affects younger users.
- The Product Solution: A gamified "Financial Command Center" featuring a centralized debt management hub (DTI metrics, waterfall payoff methods), a Stock Backtest Engine settling T+2.5 logic, and a voice-enabled AI mascot ("Vẹt Vàng") powered by Gemini & edge-tts.
- Key Tech: Next.js, React 19, Tailwind CSS v4, Better Auth (SQLite), Vercel AI SDK, Supabase PostgreSQL.
📐 WiseBIM / Revit-MCP — Model Context Protocol for BIM Automation
- The Industry Pain Point: CAD and BIM engineers spend hundreds of hours manually converting 2D architectural drawings into 3D Autodesk Revit families.
- The Product Solution: An AI-agent bridge utilizing the Model Context Protocol (MCP) to connect LLMs directly with Autodesk Revit's local C# APIs and Dynamo scripts, automating drawing classification, layer queries, and element generation.
- Key Tech: C# (Autodesk Revit API), TypeScript, Node.js, Model Context Protocol, Dynamo.
⚙️ ex5-backtest — Headless Algorithmic Trading Pipeline
- The Trader Pain Point: Quantitative traders lose valuable time manually compiling, backtesting, and ranking hundreds of different Expert Advisors (EAs) on MetaTrader 5.
- The Product Solution: A headless CLI automated pipeline that programmatically compiles, backtests, and ranks up to 70 algorithmic strategies simultaneously, generating data-driven strategy performance catalogs.
- Key Tech: PowerShell, MQL5, MetaTrader 5 CLI, Python.
Building automated tools to streamline information gathering, lead qualification, and developer operations:
| Project | Purpose & Product Thinking | Stack |
|---|---|---|
| social-lead-gen | Replaces manual prospecting by automatically scraping social leads, classifying intent via DeepSeek, and queueing targeted outreach. | Python, Playwright, DeepSeek |
| telegram-channel-analyzer | Automates sentiment and trend tracking in community groups, identifying key terms and tracking engagement spikes. | Python, NLP, Scrapers |
| business-deep-research | Custom workflow for multi-stage topic research using LLM chains with real-time grounding to produce comprehensive business briefs. | Python, Gemini, Grounding |
Experimental tools and frameworks developed to explore market inefficiencies and automate chart analysis:
| Project | Technical Overview | Stack |
|---|---|---|
| xauusd-smc-signal-engine | Analyzes gold price action using XGBoost classifiers and automated Smart Money Concepts (SMC) indicator calculations. | Python, XGBoost |
| btc-smc-rl-bot | Experimental reinforcement learning (PPO) model using Gymnasium to evaluate optimal entry points based on market structure. | Python, Gymnasium |
| xauusd-ichimoku-rl-bot | Reinforcement learning (PPO) agent integrating multi-timeframe Ichimoku elements for pattern analysis. | Python, Gymnasium |
| comarai-algo-promax | Backtesting pipeline combining SMC patterns, Moving Averages, and machine learning triggers. | Pine Script, Python |
| mql5-multisignal-dca-ccbsn | MetaTrader 5 Expert Advisor utilizing multi-signal modes, DCA grid logic, and automated parameter optimizers. | MQL5, Python |
| hcc-reader | A utility designed to parse and extract historical bar data directly from MT5's proprietary binary formats (.hcc/.hc). | Python |
📊 Additional Trading & Indicator Libraries (click to expand)
- pinescript-ict — ICT SSL Premium & Discount TradingView indicator (Pine Script v6)
- indicator-tradingview — Custom indicators collection
- moondev-agent — Trading agent orchestrations: Ichimoku + MQL5 backtest pipelines
- mql5-algo-trading-portfolio — MT5 Expert Advisor portfolio
- telegram-copy-signal — Automated trade signal copier dashboard
Languages & Frameworks
AI & Automation
Data Science & Quantitative
I actively configure advanced AI agentic systems and developer tools to streamline coding efficiency, token budgeting, and system integration:
- Token Optimization & Cost Control: Developing custom lightweight filters (e.g., RTK - Rust Token Killer) to strip redundant payload data, significantly reducing prompt context sizes and API costs.
- Context Preservation: Working with high-context memory swapping techniques to streamline complex refactoring pipelines across monolithic files.
- Open-Source Agent Research: Experimenting with autonomous, self-healing developer agent architectures to understand how to build resilient software maintenance tools.
- University of Information Technology – VNUHCM (2026–2028)
- B.Sc. in Information Technology (Distance Learning)
- BTEC HND in Business (Graduated 2024)
- Foundation in Business Analysis and Strategic Thinking
I am open to collaborations in Software Engineering, AI Integrations, and Quantitative Automation.
| 📞 Contact | |
|---|---|
| hungphamphunguyen@gmail.com | |
| GitHub | github.com/hungpixi |
| Website | phamphunguyenhung.com |
Built with passion, curiosity, and code.