Interview Agent
LiveA retrieval-augmented assistant grounded in my CV and project work. Ask it anything about my background: it retrieves relevant context, cites its sources, and streams a response.

Chivon Powers, PhD
AI/ML Engineer · Data Scientist
Agentic pipelines. Predictive modeling. Statistical experimentation. Evaluation-first, always.
What I build
LLM systems. Predictive models. Data products. Architecture through deployment.
How I think
Evaluation-first. If the system can't be measured, it's not ready to ship.
Featured build
Applied AI and ML, designed and shipped end to end.
A retrieval-augmented assistant grounded in my CV and project work. Ask it anything about my background: it retrieves relevant context, cites its sources, and streams a response.
Approach
Practical habits, not philosophy.
Evaluation before scale
I design test suites before I build production systems. If I can't measure it, I don't ship it.
Iteration over perfection
Working version first. Measure it. Fix the failure modes. Repeat.
Human oversight by design
Agentic systems need intervention points. Built in from the start, not after.
Scope discipline
Clear inputs. Clear outputs. Clear success criteria. Then code.
Builder
I build things digitally and physically. DIY carpentry and construction projects keep me honest about what I actually know how to do.
Colorado
Mountain lifestyle. Staying active, getting outside, and genuinely good air.
Healthy living
A real priority, not a weekend thing. Keeps everything else sharper.
Cats
Two of them. Entirely indifferent to my deployment schedule.
Contact
AI, ML, and data science. Reach out and we'll set a clear scope.