Human-in-the-loop
AI handles scale; humans provide context, judgment, and the final call.
AI suggests
Human reviews
Action taken
AI + Systems
Agentic workflows, human-in-the-loop interfaces, and products that make the machine's decisions legible — and reversible.
From input to impact, an agentic workflow
Feedback loop
User
Asks a question or uploads data
Research agent
Gathers relevant info from sources
Analysis agent
Identifies patterns & key themes
Synthesis agent
Generates actionable recommendations
Review & approve
Human ensures accuracy & trust
Actionable insights delivered
Featured exploration
DriftSense ✦
An agentic MLOps platform where a conversational Drift Agent orchestrates the whole model lifecycle: training, deployment, inference, and drift monitoring, with a human approving every step. Designed and built with Cursor and agentic AI.
- Conversational workflow setup via the Drift Agent
- One console for training, deploy, inference & monitoring
- Human-in-the-loop, every agent action is confirmed
- Drift monitoring that loops back into retraining
Agent experience design
→
→
Trust & transparency
Designing for clarity with confidence scores, reasoning, and source visibility.
82%High confidence
Customer volume
Recurring issue
Recent spike
Recovery & control
When AI is wrong, users need an easy way to correct and guide the system.
Incorrect insight Flag
Edit & provide context Edit
System learns & improves Learned
Designing with AI
AI is part of my process, not a replacement.
It helps me think bigger, iterate faster, and design smarter. In my work at UPS I use AI-native Figma tooling and agents to move from research to production faster, and this very portfolio was designed and built in collaboration with an AI agent (Claude).
Research
Understand problems
Prompt
Frame intent clearly
Generate
Explore options
Refine
Iterate with judgment
Design
Craft experiences
Validate
Test & improve
AI accelerates the process, but human thinking drives the direction.
My AI design toolkit
Figma AI
Ideation, layout options, content generation
ChatGPT
Research, synthesis, ideation & critique
Claude
Deep analysis, writing, reasoning & building
Cursor
Code, prototyping & automation
Perplexity
Real-time research & source discovery
Agents
End-to-end task automation & build
AI won't replace designers. Designers who understand systems, workflows, and human behavior will shape how AI integrates into everyday experiences.
Working on something AI-shaped?