DINQ is a next-generation career platform designed for the AI era. It serves AI researchers, engineers, founders, and creators by providing dynamic professional identity pages, precise talent matching, and collaboration-first networking. DINQ differentiates itself through AI-driven insights, real-time interaction, and a cloud-native architecture optimized for global scale.

As global AI talent communities are growing rapidly, DINQ needed to support increasing real-time collaboration and high engagement while delivering a consistent user experience across international markets.

From a technology standpoint, the platform had to handle traffic fluctuations caused by community activities and launches, while maintaining low-latency AI-driven matching at scale. Also, stable WebSocket connections across regions were essential to ensure fast iteration with reliable CI/CD pipelines.

DINQ set clear objectives to achieve:

● API response latency < 200ms (P95)
● WebSocket connection success rate > 99.5%
● CI/CD deployment time under 10 minutes
● Resource utilization at 60–80% through Horizontal Pod Autoscaling (HPA)

To meet these requirements, DINQ evaluated cloud providers against the following criteria:

● Cloud-native Kubernetes support
● Global infrastructure and CDN coverage
● Managed database and AI-friendly services
● Cost efficiency and operational simplicity

Alibaba Cloud emerged as the preferred choice due to its tightly integrated ecosystem and strong support for scalable, AI-ready architecture.

DINQ runs on a cloud-native microservices architecture built on Alibaba Cloud. Its core components include the following:

● Container Service for Kubernetes (ACK) hosting five core microservices, including Gateway, Server, Message, Discover, and Tool, with HPA scaling pods from 2 to 10 based on traffic demand.

● ApsaraDB RDS for PostgreSQL with the pgvector extension to support AI vector similarity search with millisecond-level response times.

● Object Storage Service (OSS) with CDN for storing and globally distributing static assets such as avatars and attachments.

● Server Load Balancer (SLB) ensures high-availability entry points for API gateways and WebSocket services.

● Qwen LLM provides the core orchestration and analysis for the advisor matching and recommendation scenario. It deeply understands and reasons over users' resumes and individualized requirements to deliver consistently high-quality, reliable recommendations.

DINQ implemented a Go and Python-based microservices platform built and deployed through GitHub Actions. Liveness and readiness probes ensure self-healing to ensure application resilience. Furthermore, HPA auto-scales services at 70% CPU or memory threshold, ensuring efficient resource utilization without compromising performance.

For data management, ApsaraDB RDS for PostgreSQL manages core transactional data with automated backups, while pgvector enables AI profile embedding and similarity search. Redis plays a critical role in caching online status, sessions, and offline messages, supporting low-latency real-time experiences.

File storage and delivery are streamlined through OSS with CDN acceleration, accessed via a unified custom domain: assets.dinq.me.

Real-time communication is another key aspect of the platform, with a WebSocket service designed to support up to 18 concurrent devices per user. Redis-based pub/sub enables cross-pod message synchronization, ensuring reliable real-time communications.

As DINQ continues its global expansion, the platform is focused on optimizing latency for users across China, the Asia-Pacific region, North America, and Europe. It aims to enable global acceleration for both static assets and dynamic APIs, as well as build multi-region disaster recovery to enhance platform availability.

Alibaba Cloud will continue to support DINQ’s next phase of growth through scalable infrastructure, cloud-native services, and AI-ready solutions, enabling global expansion, performance optimization, and accelerated innovation.