Skip to content
louay_amor~

available· MLOps Engineer · Nabeul, Tunisia

Portrait of Louay Amor

Louay Amor

I move ML projects out of notebooks and into production I make pipelines repeatable and track what matters I set up systems so you can trust your models and spot issues early

view systemsresume.pdfamor.louay20@gmail.com

01 — selected work

Systems

Six systems, three of them running in production. Three more shipped as public repositories.

Diabetic retinopathy screening, thesis project for Samaya Ophthalmology Clinic

Feb – May 2026

RetinaXAI

I turn vision models into clinical systemsI build more than a demo—every project gets services, registry, drift monitor, deploy pipeline

  • Handled 35,000+ fundus images from start to finish
  • Built grading and segmentation models with PyTorch and EfficientNet-B3
  • Explained each grade with Grad-CAM reviewed by clinicians
  • Generated RAG reports with strong grounding
  • Added 39 Prometheus metrics for live system health
macro AUC
0.867
QWK
0.724
accuracy
78.7%

Multi-tenant EdTech platform, client work

Aug – Oct 2026

Gamma Academy

I build real products, not just deploymentsFour consoles 110 agencies one database—no mess of multiple apps or logins

  • Built a modular monolith with NestJS for all consoles
  • Enabled partner branding and URLs with config not extra code
  • Used Cloudflare DNS and simple deploys for all partners
  • Kept all tech choices tracked in clear decision records
agencies
110
enrolled students
1,618
automated checks
1,005+

Partner-scoped console inside Gamma Academy

Sep – Oct 2026

ENDA Partner Portal

I make tenant-based systems simple not messy folders or loginsEach partner gets only what they need—branded login, scoped accounts, documents

  • Branded logins for each partner page
  • All accounts and documents scoped and secure
  • Shared backend for mail SMS messaging reused everywhere
  • No cross-partner leaks or confusion
consoles live
1
agencies
110
cross-tenant leaks
0

public repositories

Distributed hotel booking, DevOps platform exercise

2025

EasyBooking

I put service discovery config and packaging firstA scalable system is set up before feature fun starts

  • API gateway for smooth routing of users and reservations
  • Used Helm with Kubernetes for organized deploys
  • Live monitoring with Prometheus Grafana logs
  • CI for every service keeps code healthy
services
6
runtime
Compose
packaging
K8s + Helm

Streaming RAG architecture

2026

Reasona RAG

I build RAG pipelines where reranking tracking and telemetry come firstEverything is explainable and measured

  • Built streaming RAG with FAISS retrieval
  • Added reranking before generation for quality
  • MLflow tracks every experiment for ID and recall
  • Monitoring built in from the start
retrieval
reranked
serving
streaming
governance
MLflow

MLOps training pipeline

2024

Kidney Disease Detection

I focus on reproducibility before worrying about model scoresPipeline stats and every run is repeatable

  • Every pipeline stage tracked from raw data to model registry
  • All experiments logged with MLflow and DVC
  • Used AWS EC2 and S3 for speed and trackable artifacts
  • Consistent 85% test accuracy
test accuracy
85%
stages
DVC
artifacts
S3

02 — what i operate

Capabilities

Each row links to the system where it actually ran. Nothing listed here is aspirational.

Data & Pipelines

Modeling

LLMOps

Serving & Scale

Platform & DevOps

Observability

03 — how i work

Engineering practices

What shows up in the code, not what shows up on a slide.

  • Reproducible pipelines

    Params in versioned config, stages as idempotent units, artifacts tracked from raw data to the registry.

    → Kidney Disease Detection

  • Instrumented serving

    Health, readiness, and a metric series for everything a capacity plan will be asked about later.

    → RetinaXAI

  • Hard borders on multi-tenant data

    Tenant scope materialized as FKs, not folder structure or conventions.

    → Gamma Academy

  • Decision records

    Tradeoffs and migration paths written down in the repo, so the next engineer does not redo the reasoning.

    → Gamma Academy

  • Zero-downtime releases

    4-stage Docker builds and migrations that do not hold the deployment host together by accident.

    → Gamma Academy

  • Secured CI

    Per-service workflows, quality gates, and secrets scoped to the deploy target.

    → RetinaXAI

04 — where i've been

Timeline

Full-stack to MLOps, literature degree to deployment. One arc, not a resume dump.

  1. Full Stack Engineer · Gamma Academy

    Aug 2026 – Oct 2026

    • I built a platform for many agencies and students
    • Used NestJS Next.js PostgreSQL Prisma Redis
    • Checked everything with 1,000+ automated tests
    • Deployed with Docker no downtime

  2. MLOps Engineer, Final Year Project · RetinaXAI · Samaya Ophthalmology Clinic

    Feb 2026 – May 2026

    • I built a diabetic retinopathy screening system end to end
    • Used 35,000+ fundus images, handled all steps from data to deployment
    • Trained EfficientNet-B3 for 5-class grading, 0.724 QWK 78.7% accuracy
    • Kept errors low and explained model grades with Grad-CAM reviewed by doctors
    • Added 39 Prometheus metrics, detected drift/accuracy issues early
    • Built RAG pipeline for clinical report generation with LangChain and Mistral-7B

  3. Freelance Software Developer · Self-employed

    2021 – 2023

    • I worked with clients as a freelance developer
    • Delivered software for a range of domains and tech stacks

  4. VR Social Impact Project · WomenUpV · NETINFO Nabeul

    Feb 2021 – May 2021

    • I built a VR training app for psychological violence awareness
    • Used Unreal Engine 4, added behavioral tracking with Firebase
    • Created a vulnerability-scoring model with scikit-learn

education

  • Bachelor in Software Engineering

    2023 – 2026 · I earned my Bachelor in Software Engineering at IT Business School Nabeul

  • Bachelor in Information Systems Development

    2018 – 2021 · I completed a Bachelor in Information Systems Development at ISET Nabeul

05 — also on github

Archive

Public repositories, sorted by last push.

06 — get in touch

Contact

Most interested in systems work that someone has to keep running after the demo.

  • MLOps engineer roles
  • LLMOps platform work
  • Freelance ML infrastructure
amor.louay20@gmail.com