Noé Manwaring-Favennec

Engineering Student | CentraleSupélec & McGill
My academic foundation is built on an intensive, proof-based mathematics curriculum modelled on the French Classes Préparatoires. I am driven by a first-principles approach to learning and work, which I am applying to the field of Quantum Computing, and Quantum Error Correction (QEC) in particular.

Experience

Université de Sherbrooke - Institut Quantique
Research Intern — May 2026 to Present
Jus Mundi
Engineering Intern — June 2025 to July 2025
  • Architected an automated detection pipeline for both intrinsic and extrinsic hallucination in Jus AI, improving the trustworthiness of outputs for international arbitration professionals.
  • Implemented an 'LLM-as-a-Judge' approach to verify atomic claims against provided source texts, identifying discrepancies and generating a short justification paragraph for more transparency.
  • Authored a technical research paper detailing the detection methodology and evaluation metrics.
  • Presented the detection pipeline to the entire enterprise, demonstrating the system's capacities.
Thales
Intern — June 2024 to July 2024
  • Evaluated physical storage limits (analysing superparamagnetism in HDDs and Strontium Ferrite properties in LTO-9 tapes) to model the density and sustainability of long-term data architectures.
  • Optimised the GHG emissions calculation pipeline by developing automated scripts to replace manual data processing.

Education

CentraleSupélec & McGill
Bachelor of Global Engineering — September 2023 to Present
  • Served as founding Vice-President of IT for the Bachelor Student Association at CentraleSupélec. Designed and developed the IT infrastructure from scratch, including a centralised membership portal, an integrated payment and liability waiver system, and a custom mailer engine.
  • Held membership in the CentraleSupélec Fabrique (Maker Space).
  • Developed software for a line-tracking robot, combining a PID controller with a neural network that detects immobilisation and triggers computer vision for direction recovery during the Robotics Week Challenge.
Lycée EIB Étoile (International Bilingual School)
French Baccalaureate, with Highest Honours (Mention Très Bien) — September 2021 to June 2023
  • Completed the Advanced Mathematics curriculum (Maths Expertes).
  • Founded and led the Coding Club, encouraging students to learn programming through workshops and projects.
  • Compiled the entire philosophy course into a clean LaTeX PDF, creating a revision resource utilised by subsequent cohorts.
  • Developed the IT infrastructure for the UN Club's Model-COP27, automating participant tracking, white paper submissions, and event reminders.
  • Developed a large-scale polling platform for the school to support the feasibility of a teacher's project.
French International School of Beijing
Middle School — January 2016 to January 2020
  • Pursued the British International Section curriculum, studying Literature, History, and Geography in the native-speaker stream.
  • Gained early hands-on experience building systems such as a 3D printer, near-field communication devices, and control systems at the school FabLab.
  • Awarded the Jury's Special Prize (1st place across all categories) at the Nuit du C0de (French International School of Taipei), distinguishing ourselves as the youngest team to ever participate.

Papers

Translating low-resource agglutinative languages via RAG-LLM: A case study on Quenya Pre-print in preparation
24 January 2026
Abstract: This paper presents a Retrieval-Augmented Generation (RAG) framework designed to translate agglutinative languages, using Quenya as a case study. Given that Quenya is a low-resource language, we assume that any results we obtain will be a baseline for other highly agglutinative languages with greater resources. We propose a methodology that combines recursive suffix decomposition and pre-emptive inflection to deal with the morphological complexity of Quenya as well as its mutations, such as nasal infixation. By leveraging structured morphological retrieval alongside Large Language Models (LLMs), our framework allows for the disambiguation of unique grammatical features, including the dual number and pronouns with inclusive/exclusive distinctions, thereby providing an effective translation solution for highly inflected languages.
PDF / Read
A Framework for Phonetic Conversion of Hanja to Kanji Finalising
7 November 2025
Abstract: This document presents an empirical framework for converting Korean hanja pronunciation into their corresponding Japanese kanji onyomi (Chinese derived reading). By leveraging the historical borrowing of Chinese characters by these two languages, we propose a method to map pronunciations and disambiguate multiple readings as often as possible.
PDF / Read

Projects

BDB Vice-President of IT
Role

Served as founding Vice-President of IT for the Bachelor Student Association at CentraleSupélec. Designed and developed the IT infrastructure from scratch, including a centralised membership portal, an integrated payment and liability waiver system, and a custom mailer engine.

View Details
Surface Code Simulator
Online

An interactive surface code quantum error-correction simulator. Allowing the user to configure lattice dimensions, boundary orientation, toggle X/Z errors, and observe stabiliser excitations and logical operators.

Quantum Computing QEC Surface Codes
Open Simulator
Quenya RAG-LLM translator
Open Sourcing

A working prototype of the retrieval-augmented generation (RAG) system described in my research. It implements recursive decomposition and pre-emptive inflection to translate the highly agglutinative constructed language of Quenya into English.

RAG Morphological Parsing LLM Integration
Read Relevant Paper
See all projects

Legacy Projects

Custom Selection
800+ Users

A featured Chrome extension allowing users to override default browser selection styles with domain-specific CSS injection.

JavaScript CSS Injection
View Product Page
Edunao Filter
Open Source

Institutional platform enhancement script. Reorders courses and optimizes UI layout for CentraleSupélec students.

DOM Manipulation Userscript
View on GitHub
See all legacy projects