Post-doctoral contract/ Project assistant on the theme: "Artificial intelligence for low-resource languages".

CONTEXT & OBJECTIVE OF THE POSITION

Inalco is a unique public institution in the heart of the New Latin Quarter. Founded in 1795, it is the only public institution of higher education and research in the world to offer such a rich and recognized range of training in languages and human and social sciences, both in France and internationally, with over 100 languages and civilizations taught. 
Inalco teaches 9,000 undergraduate students. The Institute employs more than 500 staff.
The recruitment is being carried out as part of the Junior Professorship "Artificial intelligence for rare or poorly endowed languages", of which Inalco is a laureate.

The project aims to address the challenges posed by languages with limited digital resources. Many languages around the world lack the linguistic data and computational tools needed to benefit from recent advances in automatic language processing (ALP) and artificial intelligence (AI). This project explores methods for developing AI-based technologies capable of processing, researching and teaching these languages, with a focus on data scarcity, linguistic diversity and multilingual interoperability. The aim is to design robust models that can support a variety of linguistic applications, from text analysis to machine translation, while preserving language diversity and accessibility.

Part of this project is dedicated to speech processing for sparsely endowed languages, with a particular focus on automatic speech recognition (ASR) and text-to-speech (TTS) systems. Speech technologies require large annotated datasets, which are often unavailable for these languages, especially in varied dialectal contexts. The research focuses on data collection strategies, augmentation techniques and AI models capable of operating effectively in multilingual and multidialectal environments. By developing methodologies for training speech models with limited resources, this project contributes to the advancement of AI applications for spoken languages, improving their use in educational, cultural and technological fields.
We are looking for a postdoctoral researcher specializing in automatic speech processing, with methods from automatic language processing (ALP) and artificial intelligence (AI), applied to sparsely endowed languages. 
The successful candidate will work within the framework of the Junior Professorship "Artificial Intelligence for Poorly Endowed Languages", to advance research on speech data processing, with a particular focus on multidialectal challenges and code-switching scenarios.

RESPONSIBILITIES & TASKS

The recruited researcher will work in close collaboration with the holder of the Chair "Artificial intelligence for rare or poorly endowed languages" as well as with INALCO's ERTIM team (Équipe de Recherche Textes, Informatique, Multilinguisme). ERTIM website: https://www.inalco.fr/ertim

The postdoctoral researcher will focus on:

• Speech processing for rare or poorly endowed languages, addressing challenges related to automatic speech recognition (ASR) and text-to-speech (TTS) systems.
• The development of robust models capable of handling dialectal variation and linguistic continua.
• The exploration of data collection, annotation and augmentation methods to improve model performance in low-resource contexts.
• The experimentation of AI techniques to optimize speech technology in environments with high linguistic variability and few available resources.
• The possibility of developing pedagogical tools for underrepresented languages based on AI technologies.

The scientific activities of the post-doc will concern:

• Participation, support and collaboration in the activities of the Chair "Artificial Intelligence for Under-resourced Languages".
• Organization of a scientific event (study day in May 2026, workshops) in connection with the research project and the activities of the Chair.
• Regular participation in the scientific activities of ERTIM.
• Organization of a research field for oral data collection in a poorly endowed language (optional).
• Editing and publication of one or two articles in peer-reviewed scientific journals.

REQUIRED qualifications & SKILLS

• Doctorate in computational linguistics, NLP, machine learning or related field, obtained after 2022.
• In-depth experience in automatic speech recognition (ASR), speech synthesis (TTS) and speech processing in general.
• Familiarity with modeling sparsely endowed languages and the challenges of linguistic diversity.
• Experience in working with a sparsely endowed language.
• Programming skills in Python and proficiency in deep learning frameworks (e.g. PyTorch, LLMs language models).
• Excellent scientific writing skills.

HR information

• Type of contract: Post-doctoral for 12 months
• Gross remuneration: €2,500 monthly
• Full-time: 38h45 weekly
• 54 days paid annual leave including 2 mandatory closed periods (3 weeks in summer and 1 week at Christmas);
• All positions at Inalco are open to people with disabilities;
• Location of position: ERTIM, INALCO, 2 rue de Lille, 75007, Paris;
• Desired starting date: as of June 16, 2025.

Application file consisting of:

• A copy of doctoral diploma or certificate of completion.
• A cover letter explaining their interest and suitability for the position.
• A detailed CV with list of publications.
• A summary of the research project (two pages maximum).

The application must be returned in electronic form by May 15, 2025 to the following addresses:
Mrs Valentina Fedchenko, Junior Professorship: valentina.fedchenko@inalco.fr and copy to drh-recrutement@inalco.fr