EMBTalent is a comprehensive, AI-first talent acquisition platform I designed and built at EMB Global. The product serves three distinct user types — enterprise clients, talent partners, and internal admins — each with their own dedicated dashboard and workflow.
On the AI side, I integrated the Claude API (via Anthropic) and OpenRouter SDK to power two core features: (1) an intelligent Job Description parser that extracts structured requirements from unstructured text and maps them to candidate attributes, and (2) a GPT-powered BRD (Business Requirement Document) generator that lets clients describe their hiring needs in natural language and receive structured, formatted requirement documents automatically.
Architecturally, I chose Next.js 14 with the App Router for SSR/SSG flexibility, TypeScript throughout for type safety, and TanStack Query (React Query) for server state management and real-time data synchronization across the multi-role dashboards. I used Shadcn UI as the component foundation and Zod for runtime schema validation on all API boundaries.
I led a team of 6 frontend engineers, established the monorepo structure, set up ESLint and Prettier standards, created a shared component library, and introduced PR templates and code review processes that reduced review cycle time by 30%. The platform went from zero to serving 50+ enterprise clients, contributing to $1.8M+ in new revenue pipeline, and reducing average client hiring turnaround by 90%. Lighthouse scores are consistently 95+ across all routes through code splitting, lazy loading, and image optimization strategies
Responsibilities
Architected the entire frontend from scratch using Next.js 14, TypeScript, TanStack Query, and Shadcn UI
Integrated Claude API and OpenRouter SDK for AI-powered JD parsing and BRD generation features
Designed and built three separate role-based dashboards: Client Portal, Partner Dashboard, Admin Panel
Implemented enterprise-grade RBAC (Role-Based Access Control) for multi-tenant data isolation
Led team of 6 frontend engineers — code reviews, architecture decisions, sprint planning
Set up CI/CD pipeline, ESLint/Prettier standards, shared component library
Achieved and maintained 95+ Lighthouse performance scores through optimization strategies
Implemented real-time data updates using React Query with WebSocket integration