Trainee : End-to-end Information Extraction from Visual Documents (H/F)
Trainee : End-to-end Information Extraction from Visual Documents (H/F)
RéférenceBGL 48314627
- Stage
- LU-Luxembourg - Kirchberg
- GESTION DE PROJET ET TRANSFORMATION
BGL BNP Paribas is one of the largest banks in Luxembourg and part of the BNP Paribas Group.
It offers an especially wide range of financial products and bancassurance solutions to individuals, professionals, businesses and private banking clients.
In 2020, BGL BNP Paribas was named “Best Bank in Luxembourg” by Euromoney for the fifth year in a row.
As part of its development, we are looking for a:
Trainee : End-to-end Information Extraction from Visual Documents (H/F)
6 months
You must provide an internship agreement covering the entire period of the internship
Value of the position:
The goal of this internship is to participate in a project involving multimodal retrieval of data and information extraction tailored to specific business needs. Responsibilities include implementing and enhancing OCR-free information extraction models, evaluating and comparing with classical OCR-based pipelines, and assessing model hallucination and risk on real data.
Working Environment:
You will be joining the BGL BNP Paribas DataLab Team. The DataLab team is composed of 6 data scientists who develop AI solutions including Generative AI, Natural Language Processing, Automatic Speech Recognition, and Computer Vision for various banking use cases.
Key Responsibilities (Your Mission)
As an intern with the BGL BNP Paribas Datalab Team, you will be involved in the following activities:
- Implementing and enhancing OCR-free information extraction models.
- Evaluate and compare with classical OCR based pipelines.
- Evaluate model hallucination and assess risk of the model on real data.
- Conducting comprehensive evaluations of models to ensure their accuracy and reliability.
- Designing and refining deep learning architectures to improve performance and efficiency.
- Training models to achieve robust and generalizable results across various business-related applications.
Your Profile
We are seeking enthusiastic and talented individuals who are passionate about the intersection of deep learning, natural language processing (NLP) and computer vision.
To be considered for this internship, candidates should possess the following:
- Strong knowledge in model evaluation techniques, including metrics and benchmarks.
- Experience in training and tuning deep learning models.
- Background in OCR, vision models, or NLP is advantageous.
- Good skills in Python coding.
Education:
Master students in last year specializing in artificial intelligence, mathematics, or computer science.
Professional Experience:
Experience in model training and optimization, and familiarity with NLP and Computer Vision models.
Behavioral Competencies:
- Organizational skills
- Communication skills
- Ability to work independently and collaboratively within a team.
- Excellent problem-solving skills and attention to detail.
- Ability to synthesize/simplify
During the finalization of the recruitment process, the preselected candidate will be asked to provide us an extract from his/her criminal record dated less than 3 months (record N°3 for Luxembourg), according to the dispositions of the law from July 23 2016 concerning the criminal record.