About Romain
French
Native or bilingual
English
Fluent
German
Basic
Experience
- HeavisideAdministration Interface for a Cloud Marketing & AI PlatformCONSULTING AND AUDITSJune 2025 - July 2025 (1 month)Development of the main interface for a cloud solution using AI and automation to optimize marketing performance and reduce media cost inflation in a large insurance group.The mission includes:
- **Frontend**: migration from a no-code prototype to a Next.js + Firebase application, with fluid navigation and configurable components.
- **Infrastructure**: deployment via Terraform on Google Cloud Platform, ensuring industrialized and scalable management.
- **Security**: integration of enterprise SSO and implementation of security and access management rules.
- BISOMFine-tuning of CamemBERT & Whisper Models for Medical Dialogue AnalysisMEDICALMay 2025 - May 2025For BISOM, I led a project focused on the automatic processing of French medical dialogues, combining speech recognition and medical entity extraction.
Entity Extraction with CamemBERT
I developed a Named Entity Recognition (NER) pipeline on transcriptions of emergency calls and clinical notes, targeting sensitive entities such as name, age, address, phone number, and medical history.- Fine-tuning CamemBERT on an annotated medical dataset
- Use of Hugging Face Transformers
- Deployment of the model within their Hugging Face organization
- Production deployment via Inference Endpoints
- Inference access via API, usable for demonstrations or product integrations
- Integration of a fuzzy matching mechanism to improve robustness
Speech Recognition with Whisper on Android
In parallel, I explored the integration of Whisper.cpp into an Android application for offline transcription.- Tests performed with Whisper small converted to GGML format,
- Native Android app in Kotlin, integrating Whisper.cpp inference via JNI
- HeavisideCreation of an HR SaaSCONSULTING AND AUDITSFebruary 2025 - Today (1 year and 4 months)I am developing an HR SaaS with Heaviside that combines AI, data science, and web technologies.The platform allows HR teams to generate personalized reports, score candidates using an approach combining AI and statistical models, with an anonymization process integrated into the scoring.Stack used: Next.js, shadcn/ui, Tailwind, Firebase, GCP, Terraform, Gemini.
Reviews
Recommendations
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- M2 Probability and FinanceUniversité Pierre et Marie Curie2012
- M2 Applied MathematicsUniversité de Lille2011