KEN SANG TANG@Benny-Tang
4
Events attended
4
Submissions made
Malaysia
5+ years of experience
About me
An experienced freelance System Integrator specializing in the design, implementation, and management of complex IT solutions that integrate diverse systems and applications. Proven ability to translate business requirements into scalable, efficient, and innovative technical solutions across various industries. Adopts a strategic outsourcing model, collaborating with a network of specialized experts to deliver high-quality outcomes. This approach enhances efficiency, ensures access to advanced expertise, and enables agile, scalable project execution. Core strengths include solution architecture, project management, and optimizing business processes through technology. This combined expertise in system integration and strategic collaboration supports effective risk management and consistently delivers robust, future-ready solutions.
🤓 Latest Submissions
SEA Property Pulse
SEA Property Pulse is a live web intelligence system that continuously extracts real-world signals from the open web and surfaces them as actionable market intelligence for real estate agencies across Southeast Asia. Instead of generic lead lists, it tells agencies what is actually happening on the ground — motivated sellers pricing down urgently in Cyberjaya, expats relocating to Singapore from Hong Kong, a new MRT line pushing property values in Johor Bahru — all detected automatically from live web data before it appears in any MLS or listing platform. The system runs three signal pipelines: Seller Stress Detection (price reductions, distressed listings, urgency language), Buyer Intent Discovery (relocation discussions, expat forums, life-event triggers), and Neighborhood Momentum (infrastructure news, new developments, hiring growth). Each signal is classified by Claude AI with a confidence score and urgency rating, then surfaced on a clean agency-facing dashboard covering six markets: Malaysia, Singapore, Thailand, Indonesia, Philippines, and Vietnam. Built solo in 5 days using 4 Bright Data tools - SERP API, Web Unlocker, Datasets, MCP server for live web data extraction, Anthropic Claude API for signal classification and insight generation, Python FastAPI for the backend, and vanilla HTML/JS for the frontend. Deployed on GitHub Pages.
31 May 2026
Aegis Pro v1 — Autonomous Supply Chain Crisis AI
Aegis Pro is an autonomous supply chain crisis management system built on Google Gemini 3.0 Flash and Veea's Lobster Trap edge infrastructure. When a disruption occurs — a geopolitical event, a port closure, a shipping lane blockade — Aegis detects the signal from live marine traffic sources, performs deep geopolitical root-cause analysis, runs an ARIMA and XGBoost ensemble forecast, simulates three probability-weighted scenarios, and uses Gemini 3.0 Flash multi-turn reasoning to generate prioritised mitigation actions — each with a quantified USD savings estimate. The system then dispatches multi-channel alerts via Slack, email, and Speechmatics voice TTS, and triggers autonomous ERP and logistics workflows — all within 60 seconds of the initial signal. A new Synthesis Agent cross-validates all seven agent outputs for consistency before anything executes, ensuring auditability and enterprise compliance from day one. In our Hormuz disruption demo scenario, Aegis identifies $1.65 million in recoverable value against a $1.7 million do-nothing cost — a 97% cost avoidance rate. Enterprise clients on a $100,000 annual licence generate $3.3 million in savings from just two disruption events per year — a 33x ROI. Aegis Pro is open-source under the MIT Licence and built entirely on Google Colab, making it accessible to any team with a Gemini API key. It is production-deployable to Veea Lobster Trap edge nodes for air-gapped environments including ports, vessels, and warehouses.
19 May 2026
Aegis Pro v2 — Shield Against Chaos
Aegis Pro is an autonomous enterprise supply chain crisis management system built for the AI Agent Olympics — Milan AI Week 2026. It deploys a 7-agent AI swarm that continuously monitors live maritime shipping intelligence from gCaptain and TradeWinds, real-time commodity market signals via Kraken, and geopolitical risk factors — then autonomously delivers a complete crisis response in under 2 seconds. The pipeline runs Signal Agent for threat detection, Intelligence Agent for geopolitical root cause analysis, a GPU-accelerated ARIMA and XGBoost hybrid Forecast Agent for 14-day price and delay projections, Simulation Agent for 3-scenario planning, Decision Agent for ranked action recommendations with estimated savings, Alert Agent for multi-channel notifications, and Execution Agent for autonomous ERP workflow triggers. Built on Featherless Llama-3.1-70B as the primary LLM with Mistral-7B fallback, deployed on Vultr serverless infrastructure, and connected to Kraken public market data. A credit-protected LIVE_MODE flag ensures zero API burn during development — flipped only on submission day. Google Colab with T4 GPU serves as emergency backup. Average savings per prevented crisis: $1.7 million.
19 May 2026
Aegis — Autonomous Crisis Management AI
THE PROBLEM: Global supply chain disruptions cost enterprises $1.5 trillion annually. When crises hit — oil spikes, shipping closures, geopolitical escalations — human analysts take hours to respond. By then, damage is done. WHAT AEGIS DOES: Aegis is a fully autonomous 7-agent AI system that monitors global risk signals, scrapes live maritime intelligence from gCaptain and TradeWinds, predicts disruption impact using AMD-accelerated ML, and executes crisis response — in 2 seconds. THE 7-AGENT PIPELINE: 1. Signal Agent — detects anomalies + scrapes live MarineTraffic headlines 2. Intelligence Agent — interprets signals using Groq LLaMA 3.3 70B 3. Forecast Agent — ARIMA + XGBoost GPU-accelerated on AMD MI300X 4. Simulation Agent — 3 weighted what-if scenarios with cost estimates 5. Decision Agent — ranked action plan with dollar savings and timelines 6. Alert Agent — real-time Slack, email and dashboard notifications 7. Execution Agent — triggers enterprise workflows autonomously AMD MI300X: XGBoost via ROCm CUDA compatibility — 8-15x faster than CPU. Deployed on AMD Developer Cloud, 4 Gunicorn workers, unlimited concurrent users. Auto-monitors every 5 minutes 24/7. REAL-WORLD: Built during the live Strait of Hormuz crisis. Real gCaptain headline scraped: "Trump's Hormuz Blockade Deepened A Historic Shipping Crisis." Aegis classified CRITICAL at 95% confidence in 2 seconds. DEMO: Oil forecast +25% in 14 days. Delay risk 45.8%. Cost impact +13.1%. Savings: $1.7M. Pipeline: 2 seconds. BUSINESS VALUE: Targets logistics directors and supply chain VPs at $50M+ procurement companies. ROI after one prevented crisis per year at $2,000/month. Customers: maritime insurers, commodity traders, manufacturers with Hormuz exposure. COMPETITIVE EDGE: Unlike Resilinc or Everstream using human analyst networks — Aegis is fully autonomous. No analysts. No waiting. 2-second response. Moat is speed, autonomy and live data.
10 May 2026
