Data Engineer at Kpler (Singapore).
- π Currently building: agentic RAG and multi-agent support systems
- π± Learning: [e.g. evals for LLM agents, LangGraph deployment]
- π¬ Ask me about: LangGraph, RAG, Airflow, Debezium / CDC, ArgoCD, Spark
- π« Reach me: [https://www.linkedin.com/in/harikrishnan-rajkumar/] Β· [email]
Languages
AI / LLM
Data engineering
Infra / DevOps
π€ CORA: Customer Operations & Resolution Agent
Multi-agent customer support system for e-commerce. A triage agent reads the customer's message and routes it to specialist agents for refunds, warranty/service and shipping, resolving tickets end to end with minimal human intervention.
- Multi-agent orchestration with LangGraph and Claude, traced in LangSmith
- FastAPI backend + Streamlit chat UI, PostgreSQL for customers/orders
- Dockerized and deployable to Render
LangGraph Anthropic Claude FastAPI Streamlit PostgreSQL Docker
π¦ issue-classifier: Secure support-ticket routing for a digital bank
Takes free-text customer issues ("my card was declined", "block my lost card") and routes each one to the owning internal team, with guardrails built in.
- PII redaction with Microsoft Presidio before any text reaches the LLM
- Prompt-injection screening; fails safe to human review instead of guessing
- Team taxonomy derived by clustering the 77 intents of the banking77 dataset into 6 teams
- Tested with pytest, pre-commit and CI
LangGraph Presidio LLM guardrails FastAPI pytest
π newbie: Agentic RAG onboarding assistant
A chatbot that helps new employees get answers from the company handbook (demoed on the public GitLab handbook).
- Hybrid search (BM25 + vector) over a Weaviate vector store
- CLI to backfill and query documents; graph debuggable in LangGraph Studio
- FastAPI backend + Streamlit chat UI
LangGraph RAG Weaviate FastAPI Streamlit



