Machine Learning Engineer

**Who are we? **:Mantu is an independent international consulting player, founded in 2007. Bringing together expert and complementary brands, Mantu stands out for the breadth of its spectrum, responding to all business transformation challenges.

Its activities are divided into four practices:
Leadership & Advocacy, Technology, Digital Marketing & Experience, Total Talent Management.

A wide range of skills, all serving a single mission: connecting and powering companies with leading teams and technology to succeed faster and sustainably.

From its headquarters in Geneva, Switzerland, Mantu relies on a community of 11,500 talented people in more than 60 countries on 5 continents and has a turnover of 1billion euros.
**Job description**:

- Innovation and R&D are at the heart of our business, which led to the creation of the Mantu Innovation Lab in 2018.

This lab enables us to stay at the forefront of new technologies and helps us to play an active role in the development and deployment of new services and products tailored to the changing reality of the market.

By combining our skills with the power of technology, we can support businesses in all their transformation requirements and increase their efficiency.
Our mission is to help our customers evolve their markets and find new opportunities for growth, to make them aware of the importance of these disruptive technologies and to help them make the most of these new tools.

***Proposed internship topic: Development of robust models to deal with disturbed signals in intelligent buildings**:
**Context**

Intelligent building management systems rely on sensors which, when faulty, can transmit erroneous data, disrupting predictive algorithms. For example, a faulty temperature sensor can lead to excessive energy consumption and reduce occupant comfort. Predictive maintenance, while useful, does not always offer instant correction, leaving automated systems to deal with corrupted data.

To overcome these shortcomings, it is necessary to develop robust models capable of maintaining their performance in the presence of noise in the data. These models would complement predictive maintenance, enabling more efficient management of buildings, even when sensors fail.

**Aim of the course**

The aim of the internship is to design and evaluate robust prediction models adapted to the specific characteristics of intelligent buildings, characterised by heterogeneous and often disturbed data. The objectives include:

- Analysing the impact of noisy data on predictions.
- Developing models capable of tolerating these disturbances.
- Proposing a hybrid approach combining model robustness and predictive maintenance.

**Profile required**
- Final-year engineering student or Master's in computer science, data science or artificial intelligence. Skills required:

- Knowledge of machine learning.
- Proficiency in Python and data science libraries (Numpy, Pandas, Tensorflow, Pytorch, etc.).

**Equal opportunity**: