ExtremeMind Studio

Overview

A full-stack intelligent merchant traffic analytics platform built with ASP.NET Core, Vue 3, UniApp, MySQL, and Redis. It processes edge-device traffic data, delivers real-time customer analytics, and integrates rule-based and LLM-powered business recommendations across merchant, sales, and administration applications.

Core Features

• ASP.NET Core API Architecture – RESTful backend built with C# and .NET 8, using layered controllers, services, repositories, DTO validation, dependency injection, centralized exception handling, and environment-based configuration
• High-Throughput Edge Data Pipeline – Node.js and Express ingestion service receives edge AI device events, validates payloads, pushes tasks into Redis queues, and asynchronously persists aggregated traffic data to MySQL
• Redis Caching & Asynchronous Processing – Redis provides real-time dashboard caching, device-to-merchant mapping, queue-based workload buffering, and fast access to frequently requested traffic summaries
• Data Persistence & Analytics – MySQL and Entity Framework Core manage merchant, traffic, customer-profile, commission, and AI-advice data, with indexed time-series queries and idempotent upsert operations for reliable aggregation
• AI & Automation Integration – Integrates OpenAI-compatible LLM APIs through configurable provider adapters, scheduled background services, prompt templates, usage limits, cost tracking, and a deterministic rule engine for automated recommendations
• Multi-Client Frontend Architecture – Vue 3, Vite, Element Plus, Pinia, and Axios power the administration portal, while UniApp provides a cross-platform WeChat Mini Program for merchants and sales representatives
• Secure Role-Based Authentication – JWT Bearer authentication and role-based authorization protect separate merchant, salesperson, and administrator APIs, with password management, account approval, and permission-controlled features
• Reporting, Uploads & Operational Tooling – Supports Excel and PDF report generation, multipart image uploads, structured logging, API error monitoring, CORS configuration, .env-based deployment settings, and Docker-ready service packaging

Overview

A serverless, cloud-native real-time crypto monitoring and signal ingestion platform that turns social posts into structured crypto signals, with continuous workflow automation and a CDN-hosted dashboard. Built with TypeScript + AWS CDK for infrastructure-as-code, n8n (Docker on EC2) for always-on ingestion pipelines, API Gateway + Lambda for validated ingestion/query APIs, DynamoDB single-table design for scalable storage and pagination, and a lightweight React dashboard served via S3 + CloudFront.

Core Features

• Real-time Social Signal Ingestion – Designed and implemented a full-stack, real-time crypto monitoring and automated trading system that analyzes tweets from major accounts on X (Twitter).
• Always-on Workflows with n8n (Docker on EC2) – Deployed n8n on an AWS EC2 instance using Docker to run continuous ingestion workflows, enabling real-time monitoring of data.
• Streaming API + Webhooks + Gemini NLP – Integrated the Twitter Streaming API and n8n Webhooks to continuously ingest tweets, with Google Gemini extracting crypto-related keywords through a custom NLP prompt.
• Validated Ingestion & Query API – Built a validated ingestion API using API Gateway + Lambda, enforcing required fields and writing suitable records into a single DynamoDB table.
• On-chain Token Analysis & Signal Filtering – Implemented on-chain token analysis using Binance Smart Chain (via DexScreener API), validating token market cap and liquidity against predefined thresholds before generating trading signals.
• IaC, Least-Privilege IAM & Automated Deployments – Provisioned and wired infrastructure-as-code with AWS CDK, including least-privilege IAM and automated dashboard deployment.
• CDN-hosted Dashboard with Pagination – Delivered a CDN-hosted dashboard (S3 + CloudFront), supports search/filtering, and incremental “Load more” pagination.

Overview

A cutting-edge AI-powered interview platform built with Azure cloud-native architecture, featuring GPU-accelerated Azure GPT-OSS inference, real-time virtual human interaction, and enterprise-grade speech processing capabilities.

Core Features

• AI-Powered Interview Engine – Real-time question generation and scoring using Azure GPT-OSS with GPU acceleration, enabling adaptive follow-up questions and intelligent evaluation workflows
• Photorealistic Virtual Interviewers – D-ID Clips Streams integration with WebRTC for ultra-low latency video streaming, providing life-like avatars with synchronized speech and expressions
• Advanced Speech Processing – Azure Cognitive Services integration for continuous Speech-to-Text (STT) with silence detection, and Text-to-Speech (TTS) enhanced with viseme-driven facial animation
• Flexible Multi-LLM Architecture – Supports GPT-OSS, vLLM, Ollama, and Mock models with automated load balancing and fallback mechanisms for maximum reliability
• Real-Time Communication Layer – WebRTC-based peer connection with ICE/STUN/TURN support, ensuring seamless interviewer–candidate interaction across diverse network conditions
• Cloud-Native Scalability – Fully containerized architecture with .NET 8 Web API, React + TypeScript frontend, and Azure-managed services for enterprise-grade availability and performance