Shailly Gadhiya — Senior UX Designer

Selected Work

Projects I'm proud of

B2B Digital Platform
End-to-end user journey

How do you design a parts shop for workshop mechanics who can't afford downtime? I built MAN's first e-commerce platform from zero — shipped across 4 European markets.

Read case study

MAN Genuine Parts & Accessories

Automotive UX
Vehicle marketplace platform

Redesigning Germany's largest automotive marketplace — making it easier for millions of drivers to find, compare and buy their next vehicle.

Read case study

ADAC Fahrzeugwelt

UX Research & Strategy
Luxury Travel

What does it feel like to explore Antarctica? I designed an emotion-led experience for ultra-luxury cruise travellers — where every interaction had to feel as rare as the destination.

Read case study

Wanderlust — Silversea Cruises

User Research
UX Benchmark Framework

How do you know if your design is actually better than the competition? I built a benchmarking framework that turns qualitative UX into measurable, comparable metrics.

Read case study

Design Metrics for UX Benchmark

AI-Assisted UX
Data classification tool

Classifying thousands of data entries manually is brutal. I designed an AI-assisted workflow that lets analysts bulk-classify at scale — without losing accuracy or control.

Coming soon

Bulk Classification of Data

Water AI Assistant
Politecnico di Milano

A conversational AI assistant that helps people understand and reduce their water consumption — designed at Politecnico di Milano with behaviour change at the core.

Coming soon

Linda — Water Assistant (Politecnico di Milano)

AI in Practice

Not a framework. Real work.

Four moments where AI changed how I worked — each one from a shipped project, each one with a decision only a designer could make.

MAN · Research Synthesis

4 interviews → 12 themes in 1 hour

After user sessions in Köln, I used Claude to cluster sticky notes and surface behavioural patterns across all 4 interviews. What would have taken a full day of affinity mapping took an hour. I then manually validated every cluster — and cut 3 that didn't hold up — before they shaped the UX brief.

Claude Lookback Maze

AI clustered · I decided what stayed

MAN · Catalog Intelligence

2,400+ SKUs classified from raw PDFs

MAN's parts catalog lived in dense, unstructured supplier PDFs. I used AI to parse, classify and group every SKU by category, vehicle compatibility and part type — turning weeks of manual taxonomy work into a single afternoon session. The structured output directly shaped how I designed the navigation and filter system.

Claude PDF extraction

Weeks of taxonomy → one session

MAN · Wireframing

5 layout variants explored in a morning

Using the Claude plugin inside Figma, I generated early wireframe variants for the Parts PDP, category pages and VIN search flow — exploring directions that would have taken two full days manually. I picked one, threw four away, and spent the time saved on usability testing instead. The testing found the issues. Not the AI.

Figma Claude plugin

2 days → half a morning

MAN · RAG Recommendation UX

Designing the AI failure state nobody built

When a part is discontinued, the platform uses RAG to surface compatible alternatives ranked by fit confidence. I designed not just the happy path — but the honest failure state: if the model can't generate a reliable match, it says so clearly rather than surfacing a low-confidence result. I set <0.5% hallucination rate as a design requirement, not just an engineering metric.

RAG architecture Confidence UX Explainable AI

Target: <0.5% hallucination rate

Team Culture · From Speed to Quality

Shifting the team from "AI for efficiency" to "AI as a quality partner"

Early on, the team used AI mostly to go faster. I pushed to reframe it: AI as a collaborator for raising design quality, not just cutting time. That shift changed how we reviewed work. We started using Claude to critique our own designs before presenting them — surfacing UX issues earlier and making design reviews sharper.

Claude Figma Design critique

AI as quality partner — not just a time-saver

Team Leadership · Building Capability

Built internal AI tooling to raise the whole team's output

Beyond my own workflow, I created internal tooling and enhanced existing platforms to support faster design execution and higher quality outcomes across Design, Engineering, and the company as a whole. This included a critique-assistant Figma plugin that empowers junior designers and surfaces UX issues earlier in the process — before they reach stakeholder review.

Figma plugin Claude ChatGPT Internal tooling

Junior designers empowered · Issues caught earlier

AI doesn't design. It removes the work that slows designers down — so we can spend more time on the decisions only humans can make.

Projects I'm proud of

Not a framework. Real work.

4 interviews → 12 themes in 1 hour

2,400+ SKUs classified from raw PDFs

5 layout variants explored in a morning

Designing the AI failure state nobody built

Shifting the team from "AI for efficiency" to "AI as a quality partner"

Built internal AI tooling to raise the whole team's output

Let's build something great together

This case study is protected