Year · 2025–presentBy Ayush Niroula
A shared AI platform for WiseAdmit admissions and university-recruitment workflows. The system combines LLM assistants, university-specific agents, LangGraph workflows, conversational state and memory, RAG and semantic retrieval, program intelligence, governed tool execution, and human escalation. It also powers WhatsApp and lead workflows for qualification, conversation scoring, follow-ups, recommendations, and admissions support.
Highlights
- Architected reusable AI capabilities for admissions assistants and university agents instead of one-off prompt integrations
- Combined intent routing, qualification, conversation scoring, automated follow-ups, recommendation workflows, contextual memory, and human escalation
- Introduced governed tool boundaries so AI actions remain aligned with application permissions and business rules
- Designed the platform as a shared organizational AI layer for reusable knowledge, retrieval, tools, and agent workflows
Tech stack
OpenAI
LangGraph- LangChain
- RAG
- MCP
- Embeddings
- Vector Search
Redis
Node.js- Python / FastAPI