Architecting Intelligence
Across Systems
I am Kirubha, Experience Design Architect. 15 years in enterprise systems. 7 years specializing in AI/ML.
From making things to deciding what should exist. Shipped ML, AI, Agentic works since 2018.
My work spans multiple stages of AI maturity, designing AI systems for complex data platforms across enterprise cloud and B2B environments.
I focus on how intelligent systems reason, act, and integrate into human decision-making—prioritizing system behavior beyond surface interactions.
Making trust and intelligence legible for people
Impact Highlights
From predictive ML to multi-agent agentic systems - measured outcomes at every stage.
SHIPPED AI
6
Projects across AI maturity stages
TRUST SIGNAL
30-40%
Faster agentic AI root cause analysis
REVENUE INFLUENCE
0-1
VP Buy in secured, first AI concept launched
TIME SAVINGS
90%
Reductions in manual analysis
VIBE, VELOCITY, AUTOMATION
10+
Rapid prototying in AI tools to production
Selected Works
Enterprise AI that shipped
Real constraints. Real users. Measured outcomes. Not demos
Agentic Diagnostic & Root Cause Analysis
Founding designer for LLM-powered investigation workflows used by Snowflake and Databricks analysts. Designed agentic handoff patterns, confidence scoring interfaces, multi-hypothesis presentation, and reasoning trace formats that make AI judgment legible and trustworthy. Consolidated 4-5 fragmented diagnostic tools into a single investigation experience.
Impact
→ 30-40% reduction in RCA time (hours → minutes)
→ 4-5 tools consolidated to 1 experience
→ Reusable agentic pattern library shipped
AI-Assisted Developer Debugging
Designed AI-assisted debugging workflows for Android internal tools — conversational log explanation interfaces that let developers query complex system telemetry via natural language. Created log visualization patterns that made opaque Android telemetry interpretable, paired with embedded AI guidance directly in the debugging workflow.
Impact
→ Reduced diagnosis time for complex log analysis
→Improved bug triage quality and failure clarity
→Integrated AI into existing developer workflows
Merchant AI Recommendations & Platform
Led end-to-end experience design for the merchant platform, including a pioneering AI-powered recommendation concept and generative AI features. Established new interaction frameworks for AI-driven assortment planning. Global Tech Hackathon winner. Secured VP-level buy-in through vision-led design storytelling.
Impact
→North star AI vision adopted and built into 2 downstream product initiatives
→ Secured VP-level executive buy-in through vision-led design storytelling
→ Global Tech Hackathon AI Innovation winner
→ Established new interaction frameworks for AI-driven assortment planning
→ First AI recommendation concept for the merchant platform — 0 to 1
Process Automation & Data Analytics for Tax Tech
Designed deterministic automation systems for Amazon Fintech, a cloud-hosted low-code / no-code platform enabling finance teams to automate large-scale data workflows with governance and auditability.
As a Product Designer embedded within the product team, I led the feature update of the user access and permission model, simplifying onboarding, clarifying roles, and improving collaboration across finance teams. This effort established a strong automation foundation, reducing operational friction while enabling safe, reusable workflows at enterprise scale.
Impact
→ Enabled scalable, governed automation for finance workflows
→ Reduced operational friction across cross-team data processes
→ Improved onboarding and role clarity through redesigned access and permissions model
Spatial Geo Modeling & ML Visualization
Led UX for ML-powered platforms supporting predictive analytics for oil and gas operations. Designed interfaces displaying 100,000+ data points and 1,000+ cross-section views in a single surface. Replaced 100+ spreadsheet-driven workflows with a unified knowledge-retention system. Early agentic signal: ML surface prediction visualized as probabilistic gradients, not binary outputs
Impact
→$1.2M annual cost savings
→90% reduction in manual analysis
→100+ spreadsheets replaced by one system
Geo data models created in another application are used for spatial ML recommendation
Related Casestudies : Design System Initiative | Feature Updates for Data Visualization
Additional Case studies
AI Chatbot Assistant that merges many AI Assistants (2022)
There are about 10 services and chatbots internally that offer different capabilities and experiences to have a conversation with users. I designed for the goal to create a concierge chatbot to provide a single, seamless and personalized user experience to the visitors and drive efficiency and effectiveness and improve the overall experience. Concept design, design exploration on conversation design to improve user engagement.
Platform to analyze and simulate ML/AI models (2018)
Designed for Data scientists/geologists/reservoir engineers, who build AI models and adjust the predictions for oil & gas as accurately as possible with this tool. The experience I designed helped in data gathering, data prep and developing methods to make clean and meaningful data so that they build the ML/AI model.