P-01 / pain
Decks, not code.
Big consultancies sell strategy. You spend three months and $80k on slideware that nobody can deploy.
Booking Q1 → Q2 2026 2 engagements / quarter
[01] · For SMBs that need AI in production
Working AI in your business in 30 days — or you don't pay.
We're a small team of senior engineers — applied AI, full-stack, mobile — that designs, builds, and operates production AI for small and mid-sized businesses. No slideware. No six-month discoveries. Code in your stack — week one.
14-day prototype guarantee
Fixed-fee engagements
Code is yours, no lock-in
▸ OpenAI ▸ Anthropic ▸ Azure OpenAI ▸ CrewAI ▸ Agno ▸ LangGraph ▸ LangChain ▸ Pinecone ▸ Qdrant ▸ FastAPI ▸ Next.js ▸ Convex ▸ Supabase ▸ GCP ▸ Azure ▸ AWS ▸ Docker ▸ Vercel ▸ OpenAI ▸ Anthropic ▸ Azure OpenAI ▸ CrewAI ▸ Agno ▸ LangGraph ▸ LangChain ▸ Pinecone ▸ Qdrant ▸ FastAPI ▸ Next.js ▸ Convex ▸ Supabase ▸ GCP ▸ Azure ▸ AWS ▸ Docker ▸ Vercel
Trusted by founders & teams
Companies we've built AI for
AutopilotVA MobilityCare Verbly Ziritex Rochemère Epsidy HawkEye ThinkMed aQari Cliq Events DARIAMA AutopilotVA MobilityCare Verbly Ziritex Rochemère Epsidy HawkEye ThinkMed aQari Cliq Events DARIAMA
[02] The problem
The AI-services market is broken.
Most SMBs that try to ship AI hit one of three walls. We specifically built this to dodge all three.
P-02 / pain
POCs that never ship.
Boutique AI shops build flashy demos that fall apart the moment real customer data, latency, or cost constraints hit.
P-03 / pain
Hiring takes 6 months.
Senior AI engineers are scarce, expensive, and you need them yesterday — not after a 90-day notice period.
[03] The offer
The 30-Day Ship.
One fixed-fee engagement. One working AI system in production. Zero slideware between scoping call and shipping commit.
- 01
Week 1 — Spec & architecture
One workshop. One written spec. Cost-aware architecture. We'll tell you honestly if AI is the wrong answer.
- 02
Weeks 2–3 — Build
Working prototype in your environment by day 14. Weekly demos, no surprises, real data.
- 03
Week 4 — Ship to production
Production deploy on your cloud. Monitoring, evaluation harness, runbooks. Hand-off documented.
- 04
Days 31–60 — Operate
We stick around. Iterate on real usage, fix what breaks, train your team. Transition or extend on demand.
[04] Capabilities
What we build.
Four areas of deep, production-grade depth. Each is anchored in real shipped work — not blog-post theory.
C-01
Multi-agent systems & automations
CrewAI, Agno, LangGraph orchestration. Internal copilots, ops automations, alert-triage agents — wired into your stack with real tool-calling, not chatbots bolted on the side.
▸ Proof: Cut SOC alert triage time by 60% for an enterprise security platform.
C-02
RAG & knowledge retrieval
Vector search over your documents, knowledge bases, and case law. Pinecone, Qdrant, ChromaDB. Grounded answers with citations — built for long contexts and messy real-world data.
▸ Proof: Shipped legal RAG over case law; healthcare RAG over consultation data.
C-03
Voice & real-time AI
ASR + TTS + streaming LLM inference for conversational systems with near real-time latency. Cost-aware orchestration so you can ship without a $50k/mo OpenAI bill.
▸ Proof: Production voice LLM for adaptive language assessment.
C-04
Custom AI products, end-to-end
Full-stack engineering — Next.js / FastAPI / Convex / Supabase, edge deploy, observability. The boring plumbing AI needs to be useful in production, owned and operated.
▸ Proof: Optimized ECG model 4s → 3ms on edge hardware. Built construction PM platform end-to-end.
[05] Proof
Past work.
A sample of production AI we've shipped — for enterprise, startups, and clinics. Quantified outcomes, not slideware.
W-01 Cybersecurity
Major MSSP / Cybersecurity Provider
AI SecOps engineering · MENA, Remote
AI-powered SOC platform with multi-agent alert analysis and incident response. CrewAI + Agno orchestration, RAG over real-time threat context, hybrid Azure OpenAI + open-source models with adaptive rate limiting.
▸ Outcome
60% reduction in alert triage time
W-02 SaaS / Productivity
AutopilotVA
AI engineering & product · Los Angeles, USA
AI-powered virtual-assistant SaaS automating routine business tasks. Multi-agent CrewAI workflows for orchestration, DB retrieval, and report automation. End-to-end UX (chat, calendar, analytics).
▸ Outcome
70% of routine tasks automated · 2× efficiency vs. traditional VA
W-03 Healthcare
MobilityCare
AI engineering (freelance) · Paris, France
RAG-based voice and chat assistants for healthcare professionals. AI agents for personalized patient inquiry response. LLM pipelines for automated medical report generation from consultation data.
▸ Outcome
Production deploy in <4 months
W-04 Edtech
Verbly
Founding AI / mobile engineer · Remote
Production LLM-driven conversational system for real-time language proficiency assessment. ASR + TTS + streaming LLM inference, stateful multi-turn orchestration, usage-based cost controls.
▸ Outcome
Adaptive multi-turn voice LLM in production
W-05 Research / SaaS
Ziritex
Founding product architect · Remote
AI-native collaborative platform for scientific writing — agent-based human-AI co-authoring. Tool-using agents with structured outputs, RAG over user documents, real-time collaboration on Convex + TanStack.
▸ Outcome
Full-stack architecture, prototype to private beta
W-06 MedTech
Epsidy
AI engineering · Nancy, France
Real-time deep-learning pipeline for ECG signal segmentation, deployed on Nvidia Jetson Nano edge hardware. End-to-end production integration with cardiologist workflow.
▸ Outcome
Inference time: 4s → 3ms
[06] Voices
What clients say.
Real words from the teams we've shipped with.
◆ Legal automation
MD
Michel Dubois
Product Owner · LegalMinds
He designed and deployed NLP systems automating complex legal document generation. His ability to integrate LLMs with specific databases for precise contextual reasoning transformed our processes.
✓ End-to-end document processing
◆ AI platform
AF
Client
Founder · AutopilotX
Built our AI Virtual Assistant platform that automates business operations and customer support. What used to take hours now happens in seconds. He understood our business problem first.
✓ Hours → Seconds deployment
◆ Performance
GC
Guillaume Calmon
Engineering Lead · Epsidy
Tackled AI model optimization, deployment, and edge devices. Delivered 4s → 3ms inference time. His versatility left a significant impact on our team's projects.
✓ 4s → 3ms optimization
◆ ML expertise
ZM
Zakaria Mordi
PhD Researcher | Cybersec Engineer · Sonatrach
Deep understanding of advanced ML topics, brilliant with outside-the-box approaches. Excelled in both theoretical and practical parts—perfect for any machine learning position.
✓ Advanced model architectures
◆ Strategic thinking
MG
Mohammed Ghennai
PhD Researcher · Grenoble
What distinguishes him is the clarity he brings to problems. He was thinking seriously about multi-agent systems and production RAG infrastructure before these became buzzwords. Real deployments, measurable outcomes—that combination of strategic thinking and hands-on execution is rare.
✓ Production-grade systems
◆ Client relations
ML
Marie Lamielle
CEO · DARIAMA
Responsive, professional, and attentive with clear, effective communication. Clear deliverables and relevant solutions.
✓ Delivered on time & budget
◆ Legal automation
MD
Michel Dubois
Product Owner · LegalMinds
He designed and deployed NLP systems automating complex legal document generation. His ability to integrate LLMs with specific databases for precise contextual reasoning transformed our processes.
✓ End-to-end document processing
◆ AI platform
AF
Client
Founder · AutopilotX
Built our AI Virtual Assistant platform that automates business operations and customer support. What used to take hours now happens in seconds. He understood our business problem first.
✓ Hours → Seconds deployment
◆ Performance
GC
Guillaume Calmon
Engineering Lead · Epsidy
Tackled AI model optimization, deployment, and edge devices. Delivered 4s → 3ms inference time. His versatility left a significant impact on our team's projects.
✓ 4s → 3ms optimization
◆ ML expertise
ZM
Zakaria Mordi
PhD Researcher | Cybersec Engineer · Sonatrach
Deep understanding of advanced ML topics, brilliant with outside-the-box approaches. Excelled in both theoretical and practical parts—perfect for any machine learning position.
✓ Advanced model architectures
◆ Strategic thinking
MG
Mohammed Ghennai
PhD Researcher · Grenoble
What distinguishes him is the clarity he brings to problems. He was thinking seriously about multi-agent systems and production RAG infrastructure before these became buzzwords. Real deployments, measurable outcomes—that combination of strategic thinking and hands-on execution is rare.
✓ Production-grade systems
◆ Client relations
ML
Marie Lamielle
CEO · DARIAMA
Responsive, professional, and attentive with clear, effective communication. Clear deliverables and relevant solutions.
✓ Delivered on time & budget
FCS feedback
◆ FCS review
A&
Ali & Ameni
Clients · FCS
- Excellent work and very professional. Ils aiment bien ton travail.
- Très beau site, merci beaucoup.
✓ Quality craftsmanship
[07] Who builds it
Three engineers. One team.
Applied AI, full-stack, and mobile — covered. No PMs. No account execs. No subcontractors. You're talking to the people who write the code.
AM
T-01 · Paris, France
Amdjed
Senior AI Engineer
- 3+ years delivering production AI systems in cybersecurity, healthcare, legal, and edtech.
- Currently AI SecOps engineer at a MENA-based MSSP, shipping multi-agent SOC automation that cuts triage time by 60%.
- Built AI systems for Hawkeye (cybersecurity), AutopilotVA (SaaS automation), MobilityCare (healthcare assistants), and LegalMind (legal automation).
- Earlier work includes MobilityCare AI assistants, Epsidy ECG model optimization, and legal automation systems.
- Master's in AI from Paris-Saclay and ESI Algiers.
AS
T-02 · Algiers, Algeria
Assem
AI Engineer · Founding Product Architect
- 5+ years shipping LLM products and agentic workflows end-to-end.
- Founding product architect at Verbly, Ziritex, Rochemère.
- PhD candidate in reinforcement learning for cyber-physical systems — peer-reviewed in PeerJ Computer Science.
- GCP-certified.
YA
T-03 · Algeria, Remote
Yakoub
Full-Stack & Mobile Engineer
- Full-stack and mobile engineer specializing in scalable, SEO-friendly web and React Native apps.
- Built ThinkMed (medical AI SaaS with Darija chatbot, voice transcription, ECG analysis), Meditalk (medical social platform), and aQari (Dubai real estate mobile app).
- The engineer who ships the surface AI products run on.
[08] How it runs
The 30-day plan.
Tight loops, weekly demos, real artifacts. You always know what you're paying for — and what's shipping next.
- 01 Day 0 — 7
Week 1
Spec & architecture
- One scoping workshop
- Written, signed-off spec
- Cost-aware architecture
- Honest go / no-go call
- 02 Day 8 — 21
Weeks 2–3
Build
- Working prototype day 14
- Real data, real environments
- Weekly demos
- Eval harness scaffolded
- 03 Day 22 — 30
Week 4
Ship to production
- Production deploy on your cloud
- Monitoring + runbooks
- Full hand-off documented
- Code & IP transferred
- 04 Day 31 — 60
Optional
Operate & extend
- On-call for production issues
- Iterate on real usage
- Train your team
- Pause, extend, or hand-off
[09] Risk reversal
We carry the risk. Not you.
Two guarantees, written into the engagement. Because nobody should pay 6 figures to find out a vendor can't ship.
G-01 · binding
14-Day Prototype Guarantee
If we don't have a working prototype demoing your core use case in 14 days, we refund the engagement in full. No questions, no consultancy hours, no decks.
G-02 · binding
Code Ownership Guarantee
Every line we write is yours. Repos, models, prompts, infra-as-code. We hand over the keys at week 4 — no SaaS rent, no vendor lock-in, no 'managed services' tax.
[10] Objections, answered
Things people ask.
Q-01 How is this different from a traditional dev shop?
We're AI-native, not an agency that bolted 'AI' onto a WordPress site. CrewAI, Agno, LangGraph, RAG, voice — these are our daily tools. All three engineers ship agentic systems and full-stack production code, every week.
Q-02 What if we don't know what we want yet?
Week 1 is design. We'll do a working session, write a spec, and tell you honestly if AI is the wrong answer. Sometimes it is — and saying so is part of the value.
Q-03 Why a small senior team vs. an agency with 50 people?
Because three senior engineers shipping every day beat 50 juniors managed by 5 PMs. You're talking to — and getting the work of — the people who write the code. No layers, no slideware.
Q-04 What if our team can't maintain the system after?
We build for hand-off. Documented architecture, runbooks, eval harnesses, pair-programming sessions for your team. Optional retainer if you want us to keep operating it.
Q-05 What does it actually cost?
30-Day Ship engagements start at $10k, fixed-fee, scoped after the intro call. Bigger / multi-quarter scopes are quoted separately. No retainers required, no hourly billing surprises.
[11] · Last step
Have an AI project
that needs to actually ship?
30-minute intro call. We'll listen, ask sharp questions, and tell you straight if we can help — or who could.
Currently booking · 2 engagements / quarter