Open to Senior AI Engineer & Tech Lead roles

Hello, I'm

Elmahdi Elaazmi

Senior Full-Stack AI Engineer

Specialised in GenAI & AI agents and RAG & knowledge graphs. Passionate about edge AI & computer vision and leading AI teams.

Download CV Contact

About me

I'm a senior full-stack AI engineer and technical lead who bridges advanced ML research and production-grade software. I guide cross-functional teams that turn AI prototypes into scalable systems, from architecture to delivery. My work spans AI/ML, agentic AI, RAG, robust backends and cloud-native infrastructure. I set standards, mentor engineers and bring a founder's mindset to everything I build.

5+years shipping AI & software

15+systems in production

5industries

Research & higher ed
Fintech
Smart cities
Edtech
Energy

3countries worked in

Morocco
UAE · Dubai
UK · London

How I lead · Four rules

Grounded beats impressive

If an answer can't be traced to a source, it isn't done.

Cost is a feature

−30% LLM spend and −30% cloud cost, same quality.

Standards outlive me

Deploys, reviews and CI the whole team adopts.

Own the outcome

I've been a founder. I measure what ships.

Skills & technologies

Full-stack and AI systems, machine learning, NLP, computer vision and the tools to build and deploy them.

LangGraph
RAG
AI agents
OpenAI API
Anthropic API
LangChain
Llama 2
FAISS
Vector databases
Fine-tuning
PyTorch
Hugging Face Transformers
TensorFlow
Keras
scikit-learn
BERT & GPT models
YOLO
NVIDIA Jetson
OpenCV
Detectron2
MediaPipe
OCR
IoT pipelines
FastAPI
Node.js
Django
Flask
Express
AdonisJS
REST & WebSockets
Temporal
PostgreSQL
Redis
MongoDB
MySQL
Neo4j
Pandas
NumPy
ETL
Docker
CI/CD
Kubernetes
ArgoCD
AWS
DigitalOcean
Linux
Model retraining pipelines
React
Next.js
React Native
Streamlit
Python
TypeScript
JavaScript
SQL
C / C++
Rust
Bash

LangGraph
RAG
AI agents
OpenAI API
Anthropic API
LangChain
Llama 2
FAISS
Vector databases
Fine-tuning
PyTorch
Hugging Face Transformers
TensorFlow
Keras
scikit-learn
BERT & GPT models
YOLO
NVIDIA Jetson
OpenCV
Detectron2
MediaPipe
OCR
IoT pipelines
FastAPI
Node.js
Django
Flask
Express
AdonisJS
REST & WebSockets
Temporal
PostgreSQL
Redis
MongoDB
MySQL
Neo4j
Pandas
NumPy
ETL
Docker
CI/CD
Kubernetes
ArgoCD
AWS
DigitalOcean
Linux
Model retraining pipelines
React
Next.js
React Native
Streamlit
Python
TypeScript
JavaScript
SQL
C / C++
Rust
Bash

GenAI & LLM apps
10
- LangGraph
- RAG
- AI agents
- OpenAI API
- Anthropic API
- LangChain
- Llama 2
- FAISS
- Vector databases
- Fine-tuning
Used at Suretree · UM6P · PROD'AIR
Machine & deep learning
06
- PyTorch
- Hugging Face Transformers
- TensorFlow
- Keras
- scikit-learn
- BERT & GPT models
Used at Decenture · 1337
Computer vision & edge AI
07
- YOLO
- NVIDIA Jetson
- OpenCV
- Detectron2
- MediaPipe
- OCR
- IoT pipelines
Used at PROD'AIR
Back end & APIs
08
- FastAPI
- Node.js
- Django
- Flask
- Express
- AdonisJS
- REST & WebSockets
- Temporal
Used at UM6P · Suretree · Phoenix
Data
08
- PostgreSQL
- Redis
- MongoDB
- MySQL
- Neo4j
- Pandas
- NumPy
- ETL
Used at UM6P · Suretree · PROD'AIR
Cloud, DevOps & MLOps
08
- Docker
- CI/CD
- Kubernetes
- ArgoCD
- AWS
- DigitalOcean
- Linux
- Model retraining pipelines
Used at UM6P · Decenture · PROD'AIR
Front end & mobile
04
- React
- Next.js
- React Native
- Streamlit
Used at UM6P · Phoenix · Decenture
Languages
07
- Python
- TypeScript
- JavaScript
- SQL
- C / C++
- Rust
- Bash
Used at 1337 · 42 Network

Continuously learning and expanding my technical expertise.

Experience

A timeline of my professional journey and growth.

Oct 2025 – PresentRabat · hybrid
UM6P
Senior R&D Engineer · Technical Lead · School of Collective Intelligence
- Led a 5-person cross-functional team rebuilding Deliberatorium, a research LLM platform, into Aideator, a production system.
- Re-architected the stack (React/TypeScript, FastAPI, PostgreSQL, Redis, Docker), scaling from ~50 to 1,000+ concurrent users (20×).
- Owned the migration end to end, preserving historical data while shipping research concepts as production LLM features.
- Set Docker deploys, code review and CI practices, adopted as the default across the project.
LLM platformFull-stackLeadership
Read the case study →
Jun – Oct 2025Dubai · remote
Suretree
AI Engineer Consultant · Fintech
- Built agentic systems on LangGraph and FastAPI for document processing, automation and decision support.
- Grounded agents in a Neo4j knowledge graph and vector databases, sustaining 90% answer accuracy.
- Automated reports, newsletters and alerts into HubSpot, cutting manual reporting time by 60%.
- Balanced accuracy against cost: −30% LLM spend, and 5 proposed use cases taken into production.
AgentsRAGKnowledge graphs
Read the case study →
Nov 2024 – Apr 2025Marrakech · hybrid
PROD'AIR
AI Engineer · Smart city
- Architected real-time YOLO detection on NVIDIA Jetson for a smart-city digital twin.
- Moved inference from the cloud to the edge: +60% detection accuracy, −30% cloud cost, −50% deployment time.
- Built an energy pipeline over 50+ IoT sensors at 95% data fidelity to surface leaks and anomalies.
- Delivered Llama-2 RAG with OCR over internal archives for employee self-service.
Edge AIComputer visionIoT
Read the case study →
May 2023 – Jun 2024London · remote
Decenture
R&D Engineer · Edtech
- Built 5+ production NLP models (GPT, BERT) at 95% validation accuracy for personalised student insights.
- Engineered reproducible MLOps workflows (Git, CI/CD, PostgreSQL/Django) for retraining and automated deployment.
- Delivered full-stack features across web, mobile, internal tools and client dashboards (React, Node.js).
NLPMLOpsFull-stack
Jun 2021 – May 2023Benguerir
Phoenix Services
Co-Founder & Technical Lead · Campus startup
- Built a campus app, student portal and services platform from concept to production.
- Scaled to ~5,000 active users with campus-wide administration adoption; acquired by UM6P.
- Ran fundraising, university partnerships, hiring and product strategy alongside the engineering.
FounderProductFull-stack
Read the case study →
Aug 2021 – Mar 2022Benguerir · part-time
IRESEN
Full-Stack Developer · Green & Smart Building Park
- Delivered a Django CMS website for a green-energy research park, raising digital engagement by 60%.
- Built admin dashboards for site-management efficiency and energy-data reporting.
DjangoDashboards

Education Journey

Expected Dec 2026Certificate
RNCP7 certificate, AI & Data
France compétences · 1337 · 42 Network
State-recognised in France and EU-recognised as a Master's equivalent (Bac+5).
Apr 2026Certificate
Digital Technologies Architect, AI & ML
1337 · UM6P · 42 Network
Computer-science programme (2019–26) with an AI & ML specialisation, completed to Level 22.
2017–19Diploma
Specialized Technician, Industrial Automated Systems
IFMIAC, Casablanca
Where it started: PLCs, sensors and control systems.
2017Diploma
Baccalaureate, Electrical Sciences & Technologies
Moroccan national baccalaureate
Sciences et Technologies Électriques (STE) track.

Projects

A selection of projects that demonstrate my passion.

AI & algorithms
Gomoku AI
< 0.5 s per move
Minimax with alpha-beta pruning and time-bounded iterative deepening, behind a Flask API and a React UI.
PythonMinimaxFlaskReact
View repository ↗
Full-stack
Music Room
3 platforms, live sync
Real-time collaborative music rooms on iOS, Android and web: React Native, AdonisJS, Socket.IO, Temporal.
React NativeAdonisJSSocket.IOTemporal
View repository ↗
ML from scratch
EEG brain-computer interface
65.6% over 10 subjects
CSP written from scratch plus LDA, cross-validated per subject on motor-imagery EEG.
PythonCSPLDAMNE
View repository ↗
ML from scratch
Neural network from scratch
99.1% test accuracy
Pure-NumPy layers, backpropagation and Adam, classifying breast-cancer biopsies.
PythonNumPyBackpropAdam
View repository ↗
Computer vision
Leaf-disease classifier
8 classes
ConvNeXt fine-tuned on 8 apple and grape disease classes, with OpenCV feature transforms. Team project.
PyTorchConvNeXtOpenCV
View repository ↗
AI & LLM
Semantic book search
+19% similarity
LangChain, OpenAI embeddings and FAISS over 963 books; NLTK query preprocessing.
LangChainOpenAI embeddingsFAISS
View repository ↗

Blog

All posts →

Have a hard AI problem? Let's ship it.

Next: leading AI engineering. I started in industrial automation, ran a robotics club, founded a company, then led AI teams. Next, I'm growing toward leading AI engineering: strategy, platform and people.

Based in MoroccoArabic · English · FrenchRobotics club president · 50+ members, 10+ workshops

For hiring teams

Senior Full-Stack AI Engineer or Tech Lead, remote.

Download CV LinkedIn ↗

elaazmielmahdi@gmail.com

For projects

Agents, RAG, vision, or an architecture review.

Build an AI product An agent, RAG or vision feature, from prototype to production.
Architecture review Accuracy, cost and scaling of an AI system you already run.
Fractional tech lead Standards, roadmap and hiring for a small team.

Send the problem in a few lines → 30-minute call → written proposal. I reply within 2 working days.

WhatsApp ↗GitHub ↗

+212 675 874 185

Elmahdi Elaazmi

How I lead · Four rules

Grounded beats impressive

Cost is a feature

Standards outlive me

Own the outcome

GenAI & LLM apps

Machine & deep learning

Computer vision & edge AI

Back end & APIs

Data

Cloud, DevOps & MLOps

Front end & mobile

Languages

UM6P ↗

Suretree

PROD'AIR

Decenture

Phoenix Services ↗

IRESEN ↗

RNCP7 certificate, AI & Data

Digital Technologies Architect, AI & ML

Specialized Technician, Industrial Automated Systems

Baccalaureate, Electrical Sciences & Technologies

Gomoku AI

Music Room

EEG brain-computer interface

Neural network from scratch

Leaf-disease classifier

Semantic book search

Answers that trace back to a source

Run inference on the device, not in the cloud

From research code to 1,000 users: rebuild, don't patch

Have a hard AI problem? Let's ship it.

Senior Full-Stack AI Engineer or Tech Lead, remote.

Agents, RAG, vision, or an architecture review.

UM6P

Phoenix Services

IRESEN