- Company Name
- ENERGIE RECRUTE
- Job Title
- Data Science / Computer Vision / IA Générative pour l'analyse de plans industriels F/H - orano
- Job Description
-
**Job title**
Intern – Data Scientist (Computer Vision & Generative AI for Industrial Plans Analysis)
**Role Summary**
Support the development of advanced computer‑vision and generative‑AI solutions to extract, enrich, and reason about technical information from OCR‑processed industrial plans, schematics, and reports. Conduct research and prototype multimodal models that integrate visual perception with large language models (LLMs) to improve extraction accuracy and document understanding.
**Expectations**
- Apply state‑of‑the‑art computer‑vision techniques to identify symbols, equipment, and areas of interest in PDF plans.
- Evaluate and fine‑tune multimodal LLM architectures for extracting and validating technical data.
- Build demonstrator prototypes that showcase feasibility and potential integration into existing workflows.
- Collaborate with cross‑functional teams in engineering, data science, and domain experts to align model outputs with industry standards.
**Key Responsibilities**
- Design and implement visual‑recognition pipelines for detecting technical elements in rasterized and vector PDF documents.
- Preprocess, annotate, and curate datasets for supervised and semi‑supervised training.
- Investigate multimodal inference methods that combine image embeddings with textual context derived from OCR.
- Compare and benchmark models (e.g., CNNs, Transformers, Vision‑Language models) against baseline extraction methods.
- Develop and maintain prototype codebases, documentation, and performance reports.
- Present findings and prototype results to stakeholders, translating technical insights into actionable recommendations.
**Required Skills**
- Proficiency in Python, with experience using libraries such as OpenCV, PyTorch/TensorFlow, and Hugging Face Transformers.
- Strong knowledge of computer‑vision concepts: object detection, segmentation, and document image analysis.
- Familiarity with multimodal language models (e.g., CLIP, BLIP, LLaVA) and techniques for visual‑text integration.
- Experience with OCR output handling and post‑processing (Tokenization, formatting, layout analysis).
- Solid statistics and data‑analysis foundations.
- Ability to work independently, manage project milestones, and communicate complex results clearly.
**Required Education & Certifications**
- Current enrollment or recent graduate in Computer Science, Data Science, Electrical Engineering, or related field.
- Coursework or projects involving deep learning, computer‑vision, or natural‑language processing.
- Knowledge of industrial documentation standards (e.g., CAD, PDF, DFD) is a plus.
---