Curriculum Vitae

You can download a PDF version of my CV here:

• Download CV (PDF, English)

Summary

Applied scientist / ML researcher with 9+ years of experience in LLMs & NLP, graph learning (GNNs), information extraction, and recommender systems.
Delivered an HR recruiter assistant and a job recommender with industry partners.
Published research in venues such as EMNLP, ISWC, and CIKM.
Open-source maintainer of Pyformlang.

Currently open to Paris-based industry collaborations and roles where I can turn state-of-the-art methods into reliable products.

Experience

Télécom SudParis — Associate Professor (Applied ML/AI)

07/2021 – Present · Paris, France

• Co-delivered with an HR partner a recruiter assistant and a job recommender; integrated into client CRM; demonstrated measurable lifts over baselines and reduced time-to-shortlist.
• Built a resume augmenter that infers and verifies technical skills from public activity signals; improved discovery of latent skills and downstream match precision.
• Fine-tuned and evaluated language models (Hugging Face Transformers) for temporal QA and knowledge cleaning; optimized prompts/training for latency and cost.
• Led a team of 3 PhD students; managed two industry partnerships end-to-end (scoping → delivery); collaborated closely with engineering and product stakeholders.
• Head of AI programs at Télécom SudParis and Institut Polytechnique de Paris.

Max Planck Institute for Informatics — Postdoctoral Researcher

10/2020 – 08/2021 · Saarbrücken, Germany

• Scaled information extraction over large web crawls; built QA over extracted triples.
• Fine-tuned LLMs for knowledge discovery and validation; used multimodal signals for verification.

Télécom Paris — Doctoral Candidate (Ph.D. in Computer Science)

09/2017 – 09/2020 · France

• Published in top venues focusing on information extraction, NLP, and graph-based methods.
• Created Pyformlang, a production-quality Python library for formal language manipulation that became the most used in its category.

Haufe-Umantis — Junior Data Scientist

01/2017 – 07/2017 · St. Gallen, Switzerland

• Prototyped people-analytics features and recommenders within the HR suite.
• Presented solutions to enterprise clients (including Daimler).

Selected Projects

• Explainable Job Recommendation (GNN)
  Built a heterogeneous graph from résumés, job posts, and behavioral signals; trained a GNN-based ranker with path-based explanations; outperformed strong baselines; deployed as a POC with an HR partner.

• Knowledge Generation & Cleaning with LLMs
  Used LLMs to generate and clean knowledge, improving precision vs. manual curation and reducing operational overhead.

• Financial Time-Series with Exogenous Signals
  Integrated news and sentiment features into forecasting models; improved explainability and robustness on large datasets.

Open Source

• Pyformlang — Lead maintainer
  ~55 GitHub stars; ~110k downloads/month (PyPI / pypistats).
  Widely used for formal language manipulation in Python; adopted in coursework and research.

Skills

ML & Research
LLMs, NLP, Information Extraction, Graph Neural Networks (GNNs), Recommender Systems

Frameworks & Libraries
Python, PyTorch, Hugging Face Transformers, LangChain, vector stores

Data & MLOps
Airflow, Docker, Spark, Kafka, CI/CD, testing, monitoring, SQL / NoSQL

Visualization
matplotlib, Plotly

Collaboration
Partner management, mentoring, supervising PhD and Master’s students

Education

• Ph.D. in Computer Science — Télécom Paris
  2017 – 2020

• M.Sc. in Computer Science — ETH Zürich
  2015 – 2017

• M.Eng. in Computer Science — Télécom Paris
  2013 – 2015

Languages

• French — native
• English — fluent

Contact

• 📍 Paris, France
• ✉️ julien [DOT] romero [AT] telecom-sudparis.eu
• LinkedIn: https://linkedin.com/in/romerojulien
• GitHub: https://github.com/Aunsiels
• Hugging Face: https://huggingface.co/Aunsiels