Krish Thapa
I build and test AI agents that save businesses time and cost, and communicate how they work to both technical teams and business leaders.
I'm Krish Thapa. I was born in Nepal, raised in Dallas, and grew up on the basketball court, which is still my passion.
I'm completing a B.S. in Data Science at the University of North Texas (May 2027) and have already applied it in a business setting as a data intern at Valero Energy.
In my projects I replicate systems companies actually run, so each one produces something I can reuse in a real workflow. My focus is multi-agent workflows, evaluations and judgment calls, LangGraph, SQL, and the tradeoffs behind agent design.
How I work
- Team first. I do my best work with others, and many of my strongest ideas started as a teammate's question.
- Curious by default. I ask questions early to understand the real problem before I build, which saves rework later.
- Clear communicator. I translate technical concepts for non-technical audiences, a skill I sharpened in door-to-door sales and customer service.
Modernize manual workflows
Bring AI workflow engineering to teams and industries where day-to-day operations are still slow and manual, and deliver measurable gains in time, cost and accuracy.
Evaluate before deploying
Run evaluations, make the judgment calls on when an agent should act or ask, and weigh cost, speed and accuracy tradeoffs in agentic systems.
Bridge leadership and engineering
Explain what a system does and why it pays off to executives, then walk engineers through every model, step and tradeoff, keeping both groups aligned.
What I work with
Click a skill to see which projects use it.
Systems I've built
Each one replicates something a real business runs. Click any step in a diagram to see what it does.
Where I've worked
Data roles taught me the tools. Sales and service jobs taught me the people side, which is most of the job in tech.
What I'm learning
Bachelor of Science in Data Science
Graduating May 2027 · Dean's List 2026
Amrit Godar
Meta · former Director of Data Analysis, KPMG
I meet regularly with Amrit, who has led data analysis work at KPMG and now works at Meta. Much of what I know about how real teams scope, build and ship data work comes from those conversations, and they shape which projects I choose to build.
Credentials
AI Fluency: Framework & Foundations
Anthropic's course on working with AI effectively, efficiently, ethically and safely, built around four competencies: Delegation, Description, Discernment and Diligence.
What I learned: deciding which parts of a task to hand to AI, describing them precisely, critically checking the output, and taking responsibility for what ships.
Claude 101
Anthropic's introduction to using Claude for real work, from everyday tasks to projects, files and multi-step workflows.
What I learned: how to structure requests and context so Claude produces reliable results, and how to turn one-off prompts into repeatable workflows.
Python for Data Science, AI & Development
IBM's course on Python fundamentals for data and AI work.
What I learned: Python data structures, pandas and NumPy for analysis, and working with APIs and web data. The base I built my later pipelines and agents on.
Excel Basics for Data Analysis
IBM's course on analyzing data in spreadsheets.
What I learned: cleaning and wrangling raw data, and summarizing it with pivot tables, filters and formulas. I used these directly in my Valero and D&K PowerWash reporting.
Let's talk about what your team is building.
I'm looking for full-time data and AI roles starting summer 2027, in Dallas–Fort Worth or remote.