# Zach (Sadok) Abdelli

> AI engineer and forward deployed engineer (FDE) consultant. Zach (Sadok) Abdelli embeds with supply chain, logistics, e-commerce and operations teams to design, build and run production AI agents, LLM evaluation suites, workflow automation and AI command centers.

Also known as: Sadok Abdelli, Zach Abdelli.
Website: https://sadokabdelli.com/
Contact: contact@sadokabdelli.com
LinkedIn: https://www.linkedin.com/in/sadokabdelli
X: https://x.com/sadokabdelli

## Roles
- AI Engineer: LLM applications, tool-using AI agents, retrieval (RAG), LLM evals
- Forward Deployed Engineer (FDE): embedded 6-14 week engagements from discovery to production
- AI Automation Consultant: roadmaps, build-vs-buy, model evaluation, executive readouts

## Services
- Forward deployed engineering engagements
- AI agent development (planners, evals, guardrails, human-in-the-loop)
- AI and workflow automation (LLMs, RPA, integrations)
- Supply chain and logistics AI (warehouse slotting, dispatch, ETA prediction, anomaly detection)
- Advertising automation across Meta, Google and Amazon
- Internal tools, dashboards and AI command centers
- AI consulting and strategy

## Selected case studies
- 3PL pick floor: nightly autonomous slotting agent; -31% pick travel in 12 weeks
- Regional service operator: voice + chat dispatch agent; -71% tickets to humans; live in 9 weeks
- $40M consumer brand: ads command room across Meta, Google and Amazon; +38% contribution margin

## Stack
Claude, OpenAI, LangGraph, Vercel AI SDK, n8n, Make, Pinecone, Postgres, Supabase, Snowflake, dbt, Segment, AWS, Cloudflare, Twilio, Next.js, TypeScript, Tailwind.

## Education
- Doctor of Business Administration (in progress)
- MBA, Strategy & Operations
- B.Sc. Computer Science

## Essays
- [Why most enterprise AI agents fail in week six — and the operating model that fixes it](https://sadokabdelli.com/writing/why-most-enterprise-ai-agents-fail-in-week-six.html): It happens with uncanny consistency. The pilot ships in two weeks. The demo lands. Leadership is energized.
- [Slotting is the most under-invested AI win in warehousing](https://sadokabdelli.com/writing/slotting-is-the-most-under-invested-ai-win-in-warehousing.html): Warehouse operators spend enormous energy on labor scheduling, carrier selection, and demand forecasting.
- [Build vs. buy in the agent era: a CFO-friendly framework](https://sadokabdelli.com/writing/build-vs-buy-in-the-agent-era.html): Three years ago, "build vs. buy" in enterprise software was mostly about integration effort and maintenance overhead.
- [Evals are the moat — model choice is just a vendor decision](https://sadokabdelli.com/writing/evals-are-the-moat-model-choice-is-just-a-vendor-decision.html): Every enterprise AI conversation eventually arrives at the same question: which model should we use? Teams spend weeks on model evaluations, benchmarking…