Skip to content
View harikrishnan-dev's full-sized avatar

Block or report harikrishnan-dev

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
harikrishnan-dev/README.md

Hi, I'm Harikrishnan πŸ‘‹

Data Engineer at Kpler (Singapore).

  • πŸ”­ Currently building: agentic RAG and multi-agent support systems
  • 🌱 Learning: [e.g. evals for LLM agents, LangGraph deployment]
  • πŸ’¬ Ask me about: LangGraph, RAG, Airflow, Debezium / CDC, ArgoCD, Spark
  • πŸ“« Reach me: [https://www.linkedin.com/in/harikrishnan-rajkumar/] Β· [email]

πŸ› οΈ Tech stack

Languages

Python Scala

AI / LLM

LangGraph FastAPI Streamlit

Data engineering

Apache Spark Apache Iceberg Airflow PostgreSQL

Infra / DevOps

Docker Kubernetes ArgoCD

πŸ“Š GitHub stats

Profile details

Streak

πŸš€ Featured projects

πŸ€– 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

🏦 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

πŸ“š 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

Pinned Loading

  1. cora cora Public

    customer operation and resolution agent

    Python

  2. issue-classifier issue-classifier Public

    A small AI App to classify user query and assign it to a corresponding team.

    Python

  3. SqlAgent SqlAgent Public

    A simple tool using LLM to convert natural language into a SQL Query and interpret the data

    Python

  4. newbie newbie Public

    An AI based Rag to help employees on-board easily and get more details about the company.

    Python