Sam
Senior product manager. Six years fixing growth, activation and retention problems at B2B SaaS and marketplace companies across MENA — now architecting multi-agent workflows that ship production software, spec to deployed product.
Cairo · Working across MENA, EMEA and GCC
Selected work
Orvo AI Ltd · 2026
AI brand monitoring across Arabic and English
Built the full product: data model, signal taxonomy, analytics layer, and an AI chat interface over noisy multi-source MENA data. Scraping and enrichment pipeline on Trigger.dev, verified end to end at 121 signals with zero errors. Automated reporting out to PDF and PPTX.
Orvo AI Ltd · 2026
Architected a Director → Builder → Review workflow where agents plan, build, test and deploy via Claude, Supabase, Vercel, GitHub and Linear — no manual code. The review gate exists because the first version reported successful completion on work it hadn't performed. I built isolated verification to prove real execution before anything deploys.
Crowd Analyzer · 2025
Search success rate +15%. AI iteration cycles −45%.
plain English into complex Boolean queries
Enterprise analysts were losing insight to query syntax. Shipped an LLM assistant that converts natural language into structured Boolean logic, plus a validation protocol with QA and Data Science for model releases.
Crowd Analyzer
Churn 30% → 22%. A 27% reduction.
The product worked; the request pipeline didn't. No prioritization logic, no segmentation, no SLA — so CX couldn't tell anyone when their issue would be resolved. Built a customer health scoring engine on four weighted factors, surfaced live in Metabase, and a request prioritization engine on top of it.
Crowd Analyzer
Conversion 30% → 50% in three months. A 67% lift. No discount.
The team's instinct was that the price was too high. It wasn't — it was indefensible, with no logic anchoring it to the market. Rebuilt pricing off a full competitive analysis, stress-tested against company financials, and cut trial quota burn with a Boolean query methodology for CX so trial users saw the product at its ceiling instead of burning out on noise.
Fanni
Cancellations −25% in one quarter.
Orders were failing late, when recovery was impossible. Built a risk scoring model on historical and behavioral signals that flagged high-probability cancellations early enough for Ops and CX to intervene.
Results
−60%
data costs
+40%
engagement
+70%
SLA
+35%
profit
+40%
BNPL activation
My Playground
Three products I built from scratch, alone, in my free time. No team. No budget. Just curiosity and too many late nights.
Build
Orchpad
An AI-native workspace where founders turn ideas into shipped products using agent workflows.
Learn
skim.
Drop in anything you need to learn. It builds the lesson around you.
Jobs
Jorvo
Finds the roles, scores them against your CV, and preps you to win — before anyone else applies.
Writing
Soon
Soon
Soon