Fedi Naimi — AI Systems Architect

I design autonomous
multi-agent systems
that replace entire manual workflows.

The AI is just delivery. The research underneath is the product. Most recently: CTO at Prospecter, 0 → 15 paying clients in 20 days.

See the work

Tell me your workflow

Backed by

Anthropic
NVIDIA
Microsoft
NSF I-Corps

4
Systems in production: 15
Paying clients in 20 days: 5,000+
AI interactions / customer / mo: 9
Engineers hired and led

The method

How I
engineer
products

I don't start with code. I start with the workflow that's bleeding money, and engineer backward to the system that fits it. The tool amplifies judgment. It doesn't replace it.

01
Niche
Find one narrow, expensive workflow worth automating. Not a market. A workflow.
02
Discovery
Talk to the people who feel the pain. Who has it, how often, what it costs them. No interviews, no build.
03
Workflow Map
Decompose the manual process step by step. Separate what a machine should do from what needs human judgment.
04
Architecture
Design the system before the code: orchestration, data model, guardrails. Never code-first.
05
Build & Ship
Production stack, in real hands, fast. Shipping beats polishing.
06
Fit & Scale
Measure time saved, ROI, revenue. Iterate until clients are genuinely happy. Then scale the team.

Selected work

Systems in production

Real customers. Real revenue. Every card here is a shipped system, not a demo. Flip one for the full picture.

In production

Prospecter

CTO & architect

Multi-agent outbound built on months of buying-intent research. 0 → 15 paying clients in 20 days. 6.25× more meetings booked.

Prospecter

Sep 2025 — Present

Reply rate: 7% → 20%; Reply rate
Meetings booked vs. baseline: 6.25×; Meetings booked vs. baseline
Paying clients in 20 days: 15; Paying clients in 20 days

The system

We spent months on the research first: the psychology of buying intent, the window between pain and urgency. The system watches for the moment a company has a real reason to care (a new hire, a funding round, a signal others miss) and sends then. A LangGraph-orchestrated multi-agent pipeline on Claude: signal-detection agents surface in-market buyers, enrichment agents score them against the exact ICP, and outreach agents draft messages grounded in the signal itself. Human review stays at the reply stage. NestJS coordinates the graph; PostgreSQL and Redis back the state.

What I built

Architected the platform from zero to production as CTO
Designed the intent-signal pipeline, scoring model, and multi-channel sequencing
Led technical hiring and engineering team execution
Cut 200 emails a day to 50. Fewer darts, right target, right moment

LangGraph
Claude API
NestJS
Next.js
PostgreSQL
Redis
Docker

prospecter.io

In production

Wayloom

Co-founder & lead engineer

60–90 minute paper dementia screens, digitized into a chat-based multi-agent assessment. NSF I-Corps. GVSU research partnership.

Wayloom

2025 — Present

Paper screens digitized: 60–90 min; Paper screens digitized
I-Corps customer discovery: NSF; I-Corps customer discovery
Research partnership: GVSU; Research partnership

The system

Multi-agent conversational workflows on LangGraph administer, score, and interpret standardized tests through natural chat. A digitization pipeline turns any test format into structured, programmable items. Adaptive engines replace the paper protocol and produce clinician-ready reports, with longitudinal monitoring past the one-off visit.

What I built

Designed the multi-agent assessment engine on LangGraph
Built the test digitization pipeline, any format to structured items
Ran NSF I-Corps customer discovery with the clinicians who would use it
Shipped the clinician portal and patient assessment flow

LangGraph
Next.js
FastAPI
PostgreSQL
Azure

wayloom-ai.com

In production

Parlivo

Lead engineer

One reasoning pipeline: Analyze → Strategize → Draft → Critique on Gemini 2.5 Pro. 5,000+ AI interactions per customer per month across 5 channels. Solo, in 4 months.

Parlivo

Dec 2025 — Present

AI interactions / customer / mo: 5,000+; AI interactions / customer / mo
Channels, one pipeline: 5; Channels, one pipeline
Zero to production, solo: 4 mo; Zero to production, solo

The system

A 4-stage reasoning pipeline (Analyze → Strategize → Draft → Critique) on Gemini 2.5 Pro. Each AI persona researches the contact, drafts channel-adapted messages, and manages multi-threaded conversations across Gmail, LinkedIn, WhatsApp, Instagram, and Telegram. BullMQ orchestrates jobs at scale; a FastAPI microservice serves inference.

What I built

Built the entire system solo, zero to production in 4 months
Designed the 4-stage reasoning pipeline for multi-channel messaging
Built the FastAPI microservice for AI inference workloads
Deployed production infrastructure on Azure

Gemini 2.5 Pro
NestJS
FastAPI
PostgreSQL
BullMQ
React Native
Azure

Watch the demo

In production

DrChop

Founder & architect

Trend signal to live, SEO-optimized Shopify listing in about 30 seconds. Live on the Shopify App Store.

DrChop

2026 — Present

Trend signal to live listing: ~30 sec; Trend signal to live listing
Real-time market signals scored: 5; Real-time market signals scored
On the Shopify App Store: Live; On the Shopify App Store

The system

A five-stage pipeline: trend research scores products across five real-time market signals — search demand, competitor pricing, supplier velocity, listing quality, price sensitivity. Approved products get matched to CJ suppliers, receive AI-generated SEO listings and enhanced images, and publish straight to Shopify. Multi-tenant, multi-store, live run logs on every stage.

What I built

Architected the end-to-end pipeline, research to published listing
Built the 5-signal scoring engine with automatic approval thresholds
Shipped the multi-tenant, multi-store platform with Shopify OAuth onboarding
Integrated CJ Dropshipping sourcing, Stripe billing, and SerpAPI market scans

Next.js
NestJS
PostgreSQL
Shopify API
CJ Dropshipping
Stripe
SerpAPI
Azure

drchop.app

What clients say

Shared with permission, names withheld.

Founder

B2B SaaS — shared anonymously

Fedi took us from an idea to a revenue-generating product faster than any engineer we have worked with.

About

Who I am

I'm Fedi Naimi, a Tunisian software architect based in Grand Rapids, Michigan. Engineer-entrepreneur: I design the systems AI runs on and lead the teams that ship them, starting from the business problem, not the code editor. The M.S. in Health Informatics (GVSU, 3.9) and the startup programs are the paper trail. The systems in production above are the proof.

Leadership is the through-line. Chairing IEEE student branches and a national section taught me the same thing being CTO did: the people-side of a system needs as much architecture as the software. I've run 500-person events and 9-engineer teams. The design work is identical — clear interfaces, honest feedback loops, no single point of failure.

Also built: Fenai (autonomous dropshipping agent — research to storefront, zero-touch) · Hackini (gamified learning platform — IEEE ESPRIT, GitHub & Microsoft for Startups).

A boat gets you across. A ship takes others with you. I'm building ships.

Anthropic Startup Program
NVIDIA Inception
Microsoft for Startups
NSF I-Corps
M.S. Health Informatics — GVSU, 3.9 GPA

Leadership

CTO
2025 — Present
Prospecter
Technical vision, architecture, and engineering team execution for an AI outbound platform.
Co-founder & Graduate Research Assistant
2025 — Present
Wayloom — GVSU
Clinical multi-agent assessment platform; NSF I-Corps customer discovery.
IEEE EMBS Tunisia Section Chair
2025 — Present
IEEE Engineering in Medicine & Biology Society
National leadership bridging healthcare, engineering, and research across Tunisia.
IT Team Lead & Scrum Master
Feb — Sep 2025
EasyBank (consultant)
Led the dev team shipping a fintech loan platform: architecture, sprints, delivery. 10,000+ clients onboarded.
IEEE ESPRIT Student Branch Chair
2022 — 2023
IEEE
20+ volunteers, 15+ technical workshops, membership up 40%.
IEEE ESPRIT Student Branch Vice Chair
2021 — 2022
IEEE
Co-organized conferences and managed logistics for 500+ attendee events.

Contact

Let's talk systems

Tell me about the workflow you want to automate. I reply within 24 hours, or email fedi.naimi20@gmail.com.

I design autonomous multi-agent systems that replace entire manual workflows.

How I engineer products

Niche

Discovery

Workflow Map

Architecture

Build & Ship

Fit & Scale

Systems in production

Prospecter

Wayloom

Parlivo

DrChop

Founder

Who I am

CTO

Co-founder & Graduate Research Assistant

IEEE EMBS Tunisia Section Chair

IT Team Lead & Scrum Master

IEEE ESPRIT Student Branch Chair

IEEE ESPRIT Student Branch Vice Chair

Let's talk systems

I design autonomous
multi-agent systems
that replace entire manual workflows.

How I
engineer
products