Research engineer - HPC applications analysis
Role details
Job location
Tech stack
Job description
The position will be conducted in the context of ESCRIPTIUM. ESCRIPTIUM aims to identify threats on supercomputers by accessing power consumption profilings. We are specifically interested to detect 1) usages that do not respect systems's usage guidelines, 2) perform non legal activities (e.g., bitcoin), or 3) are an issue for national security. ESCRIPTIUM is a collaboration between Inria Tadaam [1], the University of Bordeaux [2], and Denergium [3]. Denergium offers a software solution with key profilings datasets that describes different system usages. In turn, researchers from Inria and the University of Bordeaux will provide expertise on AI and modeling techniques to analyze the resulting datasets to eventually identify unusual patterns. [1] Equipe TADAAM [2] LABRI [3] DENERGIUM
Mission confiée
The engineer assignment is divided into four steps.
- Data curation and analysis. After collecting the dataset, the engineer will clean the dataset: specifically, they will normalize the different energy traces and reduce as much as possible the noise due to measurement. The end goal is to generate curated energy time series. While potentially tedious, this task is critical to ensure that any subsequent model can accurately detect any unusual behaviors.
- Unsupervised clustering. There is a wide variety of AI and signal techniques to analyze time series. The goal of the second task is to identify common groups of behaviors across the different energy traces. We will leverage the temporal aspect of the different series with feature extraction methods or alternatively directly with temporality aware distance metrics. The engineer will be in charge of implementing and testing these different approaches over the energy dataset.
- Augmenting application information. Beyond the energy measures from Denergium, we can extend our profiling capabilities to track more diverse information including performance counters or interference effects coupled with scalability. The engineer will assess how much extra information can improve the clustering of the different traces.
- Supervised / anomaly detection. Finally, the engineer will try to label the previously generated clusters. To do so, they will investigate known threads or unexpected behaviors to extract and characterize their profile.
Requirements
have excellent communication skills
- have experience in conducting research activities (i.e., read and understand research publications or implement research prototypes)
- have experience in signal processing
- be familiar with HPC system usage
- possess performance and energy monitoring expertise
- have background in AI techniques
Benefits & conditions
- Subsidized meals
- Partial reimbursement of public transport costs
- Leave: 7 weeks of annual leave + 10 extra days off due to RTT (statutory reduction in working hours) + possibility of exceptional leave (sick children, moving home, etc.)
- Possibility of teleworking and flexible organization of working hours
- Professional equipment available (videoconferencing, loan of computer equipment, etc.)
- Social, cultural and sports events and activities
- Access to vocational training
- Social security coverage
Rémunération
The gross monthly salary will be between 2765 and 3085€ (after witholding taxes) depending on you qualification and experience. Postuler à cette offre Partager
