Shipped, not promised.

Three different kinds of work, labelled for what they are: a deployed AI system, years of enterprise delivery, and a prototype you can test yourself.

Selected work, with the status stated plainly.

One deployed client delivery, one completed global programme, and one independent prototype. Each overview leads to the canonical case study for the complete record.

01Current deliveryDeployedClient delivery

InsideBoard

AI-assisted data-intake platform, enterprise automotive

A deployed pipeline that ingests messy enterprise Excel feeds and delivers clean, proven data, with the AI kept on a tight leash.

Role
Forward Deployed Engineer, InsideBoard AI
Context
May 2026 to present · Enterprise SaaS, automotive data

Publicly visible

Manual line-by-line file review replaced by an automated, proven pipeline

Read the case study
The client portal after a recorded run: 226 rows validated in 16 seconds, with all 89 expected columns present and in order. End-client identity removed.
02Completed deliveryClient delivery

Dior

DiorLink Clienteling, luxury CRM at global scale

Built and rolled out the clienteling platform that turned client-service agents into personal advisors, live across four global markets.

Role
Senior Salesforce Consultant, Service Cloud Lead and Product Owner
Context
4+ years · Luxury retail

Publicly visible

Live across USA, Japan, South Korea, and China, with multilingual UAT and training per region

Read the case study
With members of the CSC delivery team at Christian Dior Couture in Paris during the engagement.
03Prototype · LabIndependent build

JD Matcher

JD Matcher, an evidence-scored Salesforce fit prototype

An experiment I built for myself: paste a Salesforce job description, add a CV, and get a scored, evidence-linked verdict instead of a guess. Still a prototype, not a finished product.

Role
Solo builder: product framing, prompt design, and application engineering
Context
Ongoing, personal build · Salesforce hiring, career tooling

Publicly visible

Every score is evidence-linked: each matched requirement points to the specific CV text that supports it

Read the case study
The earlier Judjit extension beside a live public job page, returning a deliberately low 16/100 verdict with the main mismatch, role signals, evidence, and blocker visible. The interface has since evolved into JD Matcher.

The full timeline connects the projects.

View full experience
Work, case studies | Marwane Ait Yahia