Léonard Noth
ENFRDE
what I work with

Tools are cheap.Judgement is the skill.

Everything below is something I've shipped with, not something I've read about. Each one links to the projects it was used on.

01

AI & data

Where my degree is, and where most of my curiosity goes.

  • PyTorch

    Classifiers trained, exported to TorchScript, served in production

  • Transformers

    BERT fine-tuning for temporal and sentiment classification

  • Embeddings & retrieval

    Semantic search, similarity, recommendation over vector representations

  • LLM pipelines

    Schema-constrained calls, prompt-as-config, generation and augmentation of labelled data

  • scikit-learn

    Classical baselines, stratified splits, the metrics I report

  • spaCy & NLP

    Custom NER, sentence segmentation, rule-based tagging

02

Backend & cloud

Serverless by default, and infrastructure I can redeploy without holding my breath.

  • Python

    Lambdas, pipelines, training, tooling

  • AWS

    Lambda, DynamoDB, Step Functions, AppSync, Cognito, S3, CloudFront, SES, Athena

  • Infrastructure as code

    CDK stacks across two regions, with a deploy gate that blocks destructive diffs

  • DynamoDB

    Access-pattern-first modelling across 28 tables

  • GraphQL

    AppSync schema, resolvers, subscriptions for real-time collaboration

  • Node.js

    Services and tooling

  • PHP

    Where it started professionally, and a CRM still in daily use

  • SQL

    Relational modelling and reporting, from MySQL behind a CRM to local SQLite

03

Frontend & product

The part users actually touch, which is also what I teach.

  • React

    Two production apps sharing one component library

  • TypeScript

    End-to-end typing, generated GraphQL types

  • Next.js

    This site, and the Atelier suite

  • Tailwind CSS

    Design systems that stay consistent across apps

  • Svelte & Vue

    Client projects and the SAFE interface

  • Flutter

    Cross-platform mobile, shipped on both stores

  • HCI

    Interaction design, applied before I taught it at HEIA-FR

04

Practice

How the work gets made, not just what it's made of.

  • Agentic development

    Coding agents with repo-level instructions, written skills and hard gates

  • Docker

    Reproducible services, container-based Lambdas

  • Technical leadership

    Architecture decisions, and directing students and freelancers

  • Teaching

    HCI and semester projects at HEIA-FR, plus products built with pedagogues

  • Research

    An Innosuisse project, a master's thesis, two publications