Ayman Oukacha

$ whoami

Ayman Oukacha

Software architect rooted in backend engineering and DevOps. I design scalable cloud systems integrating Deep Learning, Generative AI and Semantic AI — specialised in Knowledge Graphs and ontology-grounded RAG architectures for high-fidelity inference. Leveraging Model Context Protocol and secure sandboxing, I bridge robust infrastructure with autonomous agentic reasoning to deliver verifiable, intelligent automation.

Get in touch GitHub LinkedIn Paris, Île-de-France, France

// impact

Over six years I've shipped production systems across organisations large and small — from a global IT consultancy to a focused product team — and the work sits at both ends of the spectrum: AI-native platforms and the plain, hard backend engineering that makes any product run.

6+

years shipping in production

5

big groups, ESNs & startups

Varied

domains — quick to adapt

AI + Core

equally fluent in both

delivered into › FinTech Public Sector Industrial IoT Workplace / PropTech Home Security
AI-native core engineering

AI-native systems

GenAI · graphs · ML · vision
  • GenAI due-diligence platform (KYC / AML / ESG)
  • Document-forensics anti-fraud pipeline
  • GraphRAG + ontology-grounded retrieval
  • Agentic workflows via MCP & tool orchestration
  • ML anomaly detection & computer vision

Platform & product engineering

backend · cloud · data · UI
  • Workplace-management SaaS (React / Spring)
  • Event-driven microservices (DDD / CQRS / Kafka)
  • Bare-metal → cloud-native AWS migration
  • Real-time dashboards & IoT analytics
  • CI/CD, Kubernetes & production observability
// measured outcomes
−50%

hallucination & factual error

+10%

RAG precision

−60%

GraphRAG batch runtime

−25%

entity duplication

// skills

Generative AI & LLM Architecture

RAG GraphRAG Prompt Engineering Fine-tuning TGI vLLM Ray

Agentic Systems & MCP

AI Agents Model Context Protocol Tool Orchestration Skills-based Agents Sandboxing

Knowledge Graphs & Semantic AI

RDF OWL SHACL SPARQL Ontology Engineering LPG-to-RDF Neo4j

Vector & Graph Databases

pgvector Qdrant Weaviate Neo4j SPARQL Endpoints

Machine Learning & Deep Learning

TensorFlow PyTorch Scikit-learn Model Serving

Computer Vision

YOLO OpenCV OCR olmOCR DocLing

AI Platform Engineering

LLM Serving GPU Orchestration Batch Pipelines Kestra Workflow Orchestration

Kubernetes & Containers

Docker Helm K8s CI/CD GitOps

Cloud Architecture

AWS EC2 S3 Lambda Terraform Cloud-native

Event-Driven Architecture

Apache Kafka CQRS DDD Messaging Queues Microservices

Backend Engineering

Spring Boot Java Python FastAPI REST API Node.js

API Security & Hardening

MicroVM Sandboxing Firecracker Auth Zero-trust API Gateway

// experience

AI Solutions Architect — GenAI, Semantic AI & Knowledge Graphs

September 2024 – Present

TALIUM · Strasbourg, Grand Est, France

Led the architecture of two AI platforms at TALIUM: a GenAI due-diligence platform for FinTech combining semantic knowledge systems and trustworthy AI, and an anti-fraud system for a public training-funding body detecting fraudulent applications via document forensics and computer vision.

  • Architected a modular, event-driven AI platform for KYC, AML, ESG and compliance workflows — combining knowledge-augmented generation, agentic retrieval and knowledge graphs.
  • Built an anti-fraud detection system — a document-forensics pipeline (OCR, signature detection on attendance sheets, multi-model computer vision with YOLO/olmOCR), XGBoost & Random Forest ensemble scoring, and an Imitation-Learning loop trained on fraud analysts’ feedback for continuous improvement.
  • Designed the semantic foundation with RDF/OWL/SHACL, ontology alignment and LPG-to-RDF translation — cutting entity duplication by 25% and raising RAG precision by +10% on internal benchmarks.
  • Reduced hallucination and factual-error rate by 50% through ontology-grounded retrieval and reasoning.
  • Implemented agentic workflows with MCP and skills-based tool orchestration for structured QA and enterprise-system interaction.
  • Optimised the GraphRAG and LLM-serving stack, cutting batch runtime by 60%.
  • Hardened security via microVM sandboxing, API hardening and AI-oriented UI/UX design.

Full Stack Engineer

September 2022 – September 2024

AREMIS Group · France

Designed and evolved Archibus, a React/Spring workplace-management platform for large enterprise clients.

  • Developed full-stack features with Spring, React and Node.js in a microservices architecture.
  • Contributed to AWS architecture and the migration from bare-metal to cloud-native, with CI/CD automation and secure database operations.
  • Worked on real-time workspace monitoring, computer vision and occupancy analytics via IoT devices.
  • Contributed to ML-based occupancy prediction and intelligent-workflow use cases.
  • Took part in technical design, client workshops, POCs, documentation and UI improvements.

Backend Developer

January 2022 – September 2022

Euro Protection Surveillance · Strasbourg, Grand Est, France

Contributed to the design, development, testing and deployment of a scalable event-driven microservices platform with Spring Boot, DDD, CQRS, Docker and Kubernetes.

  • Developed event-driven microservices with Spring Boot applying DDD and CQRS principles.
  • Supported containerisation, orchestration and deployment with Docker, Kubernetes and CI/CD pipelines.
  • Maintained service reliability and production availability via automated testing and observability.

Full Stack & AI Engineer — ML, Computer Vision & Kafka Microservices

February 2020 – January 2022

CGI · Strasbourg, Grand Est, France

Contributed to industrial solutions combining ML for IoT anomaly detection, computer vision, Kafka event-driven microservices and Angular UI/UX.

  • Worked on machine learning and computer vision for industrial process monitoring and anomaly detection.
  • Developed Spring Boot microservices and event-driven architectures for real-time workflow orchestration.
  • Applied ML on IoT and streaming data for anomaly detection and predictive monitoring.
  • Built Angular applications and dashboards for real-time visualisation.

// education & more

Education

École Nationale Supérieure d’Électricité et de Mécanique (ENSEM)

2017 – 2020

Engineering Degree — Software Engineering & Information Technology

Nancy, France

Université de Montpellier

2020 – 2021

University Diploma (DU) — Data Science & Big Data

Montpellier, France

CPGE — Lycée Mohamed V

2015 – 2017

Classes Préparatoires aux Grandes Écoles — Mathematics & Physics

Casablanca, Morocco

Certifications

AWS Certified Solutions Architect – Associate

Amazon Web Services

Neo4j Certified Professional

Neo4j

Languages

French
English
Arabic
Spanish

// contact

Let's build something verifiable.