Du führst eine Literaturrecherche zu multimodalen Daten durch und entwickelst Algorithmen zur Datenvorverarbeitung und Leistungsoptimierung.
Anforderungen
- •Currently enrolled in master's program
- •Background in machine learning
- •Experience with open-source LLMs
- •Proficiency in Python
- •Familiarity with TensorFlow
- •Advanced analytical skills
- •Problem-solving skills
- •Communication skills
- •Ability to work independently
- •Ability to work collaboratively
- •Experience with Hugging Face models
Deine Aufgaben
- •Umfassende Literaturrecherche zu multimodalen Daten durchführen
- •Framework für die Kombination von Open-Source-LLMs entwerfen und implementieren
- •Algorithmen für Datenvorverarbeitung entwickeln und testen
- •Leistungsoptimierung der Model-Integration durchführen
- •Methoden anhand von Benchmark-Datensätzen evaluieren
- •Forschungsergebnisse dokumentieren und zu internen Publikationen beitragen
- •Fortschritte und Ergebnisse der internen AI-Community präsentieren
Deine Vorteile
Motivierendes Umfeld
Fokus auf Stärken
Fitness am Arbeitsplatz
Familienfreundlichkeit
Original Beschreibung
# Internship (Master Thesis) - Data & AI
Functional area:
Research and Development
Country:
Germany
City:
Bretten
Date of posting:
Apr 22, 2025
!Your future job
**FIRST IN MIND. FIRST IN CHOICE**
We are seeking a motivated master’s student to join our research team for a thesis project focused on the combination of multiple Large Language Models (LLMs) for multimodal data ingestion. This project aims to explore and develop innovative methods for integrating various open-source LLMs to process and analyze multimodal data, including text, images, and other data types. The project would also consist of integrating the algorithm obtained within our MLOPs infrastructure.
Our current GenAI capabilities are limited to the ingestion of text only. Open-source algorithms, such as Villa, can perform semantic extraction from pictures. Additionally, when it comes to audio to text or text to audio, algorithms such as Riva can be used. Before integrating such algorithms into our production pipeline, it is essential to evaluate the accuracy and behavior of such algorithms to couple it with the required guardrails. Upon the project completion, the intention is to integrate multimodal data ingestion capabilities in our official GenAI solution enabler, used by a wide range of users across the group.
**Main responsibilities:**
* Conduct a comprehensive literature review on multimodal data ingestion and the use of LLMs.
* Design and implement a framework for combining multiple open-source LLMs to handle multimodal data.
* Develop and test algorithms for data preprocessing, model integration, and performance optimization.
* Evaluate the performance of the proposed methods using benchmark datasets.
* Document research findings and contribute to internal publications.
* Present progress and results to the internal AI community team.
**Experience requirements:**
* Currently enrolled in a master’s program in Computer Science, Data Science, Artificial Intelligence, or a related field.
* Background in machine learning, natural language processing, and deep learning.
* Experience with open-source LLMs (e.g., GPT, BERT, LLAMA) and multimodal data processing.
* Proficiency in programming languages such as Python and familiarity with machine learning frameworks (e.g., TensorFlow, PyTorch).
* Advanced analytical, problem-solving, and communication skills.
* Ability to work independently and collaboratively in a research environment.
* Experience with Hugging Face models use and contribution is a plus.
**Your benefits:**
* **Motivating environment:**
in a strong international team that enjoys a lot of freedom
* **Focus on your strengths:**
for example, with coaching and individual training programmes
* **Fitness on the job:**
through our health management with monthly changing campaigns
* **Family friendliness:**
thanks to flexitime and home office option
**Diverse by nature and inclusive by choice**
Bright ideas come from all of us. The more unique perspectives we embrace, the more innovative we are. Together we build a culture where difference is valued and we share a deep sense of purpose and belonging.
**Atlas Copco IAS GmbH**
Jana Heller
Human Resources