Skip to content
Harin Rishabh

Data Scientist · San Antonio, TX

I build AI systems, and the evals that catch them making things up.

Lately that means conversational AI and evaluation frameworks, built on a strong foundation in traditional data science, forecasting and ML engineering. This site is where the notes end up.

Harin Rishabh at a wooden table in a dimly lit restaurant, wearing a printed linen shirt, with a painting of Napoleon on the wall behind him.
Yes, that's Napoleon behind me. He reviews my pull requests.
Fig. 1: Things I know, over time2026: You are here
2017Today2031, probably

Skills: LLM applications, Conversational AI, Evaluation frameworks, RAG, Intent recognition, LLM fine-tuning, NLP, Python, ML engineering, Time series forecasting, Model governance, Snowflake, Docker, Kubernetes, CI/CD.

Latest writing

All posts →

Selected work

All projects →

LLM From Scratch

A self-directed program through the LLM stack, heading toward a GPT I build and fine-tune myself.

  • Python
  • PyTorch
  • Ollama
  • Embeddings
  • RAG
Work in progress · Private repo

Bank Churn Prediction

Seven classifiers benchmarked on customer churn, with the deployed model serving live predictions through a Streamlit dashboard.

  • Python
  • Scikit-learn
  • XGBoost
  • Pandas
  • Streamlit
  • Plotly