I’m a software developer with 10 years of experience building production systems at Amazon, now specializing in AI-powered apps, automation, and bots for businesses.
(Amazon background verifiable on LinkedIn: wzx-jason)
What I build:
• Telegram & WhatsApp bots (customer support, order tracking, notifications)
• AI web apps with LLM integrations (OpenAI, Claude, Gemini)
• Full-stack apps with Next.js + Supabase
• Automation pipelines and workflow
Every project I ship includes automated tests, CI/CD, and a handover doc — so you’re not left with a black box when the work is done. If you need something built properly the first time, let’s talk.
Skills
Experience Level
Language
Work Experience
Education
Qualifications
Industry Experience
Built a real-time Solana token monitoring bot that polls DexScreener every 5 minutes, scores tokens using a composite formula (volume 40%, liquidity 40%, rug risk 20%), and surfaces high-signal alerts via a live web dashboard. Surfaces emerging tokens within 5 minutes of meeting signal thresholds — hours before they appear on mainstream trackers. Includes a velocity gate to filter noise, SQLite dedup with 24-hour window, RugCheck API integration for safety badges, and a FastAPI backend with APScheduler. 67 automated tests (unit + integration against live APIs).
Built the backend and AI data pipeline for a pet relocation platform covering international travel regulations. Reduced research from 5+ hours across government sites to a guided 10-minute checklist. Dual-LLM pipeline (Claude + Gemini) runs in parallel, consolidates outputs with Jaccard similarity matching, and flags conflicts for admin review. Admin UI for managing live routes, step timing, and community uploads. Deployed on Vercel + Supabase with CI/CD and Sentry error tracking.
Link: portfolio-demos-weld.vercel.app
(Demo link is used as product built for a startup)
Built a real-time game discovery platform that monitors community messaging groups for game announcements. Reduced time to find a local game from 20+ minutes of manual group searching to under 60 seconds. A 3-stage LLM pipeline (regex → template → Gemini batch) parses unstructured messages and extracts structured game data — location, time, skill level. Results are served via a REST API to a web UI with browse and AI chat modes. Includes a Telegram digest bot for daily game summaries. Deployed on GCP with CI/CD, auto-rollback, monitoring alerts, and 285 automated tests.
Designed and built high-availability payment infrastructure at Amazon over 10 years. Core work on Amazon Wallet — the storage layer and backend API serving internal Amazon teams at scale. Built for reliability, low latency, and production-grade fault tolerance using Java, DynamoDB, AWS Step Functions, and SQS.
Hire a AI Developer
We have the best ai developer experts on Twine. Hire a ai developer in Singapore today.