Job Description

Bosch Research Pittsburgh would like to invite an enthusiastic and knowledgeable research intern for investigations at the intersection of Multimodal Machine Learning and Natural Language Processing. We wish to develop algorithms for sequential decision-making, on the basis of multiple sensor modalities (e.g., video frames, textual instructions, knowledge graphs), in order to satisfy such downstream tasks as natural language robot navigation, autonomous driving, or visual question-answering. We wish to integrate such algorithms into existing real-world implementations and/or address selected AI tasks in related literature.

We expect the intern to implement and evaluate various methods, inspired by their own insights, team discussion, and contemporary academic literature. Viable methods include end-to-end neural systems and hierarchical AI methods. Regardless of the chosen method(s), the intern must understand the relevant challenges, e.g., multimodal data alignment, practicalities of neural model implementation (e.g., memory usage, optimisation), model robustness and generalisability, and performance characterisation.

Together with our faculty collaborators in the School of Computer Science (SCS) at Carnegie Mellon University (CMU), we have made several key developments that we expect the prospective intern to leverage and extend. The final, key component of the internship is scientific contribution: the prospective intern will be expected to work with teammates to publish a high-quality research paper in a major conference venue (NeurIPS, ICML, ICLR, CVPR, AAAI, IJCAI, RSS, ICRA, ICCV, CoRL, ACL, NAACL, EMNLP, etc.).

TASKS:

• Perform extensive literature review, to understand the current state-of-the art
• Generate a research plan—detailing intended tasks, approaches, and evaluation metrics
• Present related work and research progress to colleagues, on a weekly basis
• Design and implement methods, according to the proposed approach
• Devise metrics and evaluate algorithms extensively (proofs, empirical baseline and ablation results, qualitative analysis)
• Summarize findings as a high-quality research paper

Qualifications

• Strong background in machine learning, with particular emphasis on neural representation learning
• Extensive experience in from-scratch deep learning model implementation, e.g., using PyTorch
• Extensive experience in software development in Python on Linux-based systems
• Extensive experience in implementing neural methods for CV and NLP, specifically
• (Preferred) Strong theoretical background in such topics as: representation learning, variational Bayes, MCMC, manifold learning, information theory, (non-) convex optimisation, kernel methods, computational complexity
• (Preferred) Experience with data analytics toolkits, such as numpy, pandas, and scikit-learn
• (Preferred) Mature researcher, with existing publication history in top conference venues

Additional Information

OTHER REQUIREMENTS

• Your degree level: pursuing doctoral degree, or current post-doctoral researcher
• Your major: Computer Science, Electrical/Computer Engineering, Statistics, or related

LOGISTICS

• Internship location: Pittsburgh, Pennsylvania, United States
• Start date: Typically, sometime between April and June
• Duration: Typically, 14 weeks (extension possible; subsequent research collaboration encouraged)

By choice, we are committed to a diverse workforce - EOE/Protected Veteran/Disabled. BOSCH is a proud supporter of STEM (Science, Technology, Engineering & Mathematics) Initiatives

• FIRST ROBOTICS (For Inspiration and Recognition of Science and Technology)
• AWIM (A World In Motion)