Data science intern
Technology domains: Python, scikit-learn, TensorFlow/Keras, Scala, Spark
Location: Paris, France
Duration: 6 months
Keywords: Internet of things (IoT), artificial intelligence (AI), data-driven real estate asset management, sensor data, smart buildings
Square Sense is a fast-growing platform that provides advanced data solutions to global real estate developers, investors and managers. The company builds AI-powered “building brains” that support the digital transformation of investment and asset management, and improve the operational and financial performance of real estate assets.
These brains provide real-time pattern interpretation of user community profiles and service performance, and enable autonomous optimization to greatly enhance the user experience for the tenants and the asset/portfolio management strategy execution such as investment, net income or ESG.
Square Sense was founded in 2017 in Paris by a multi-cultural team of talented engineers and data scientists.
Data science @ Square Sense
In Square Sense's data science team, we are developing products to understand people's behavior and building performance using data collected from sensors and systems within a building. Our models infer knowledge using data from heating, ventilation, and air conditioning (HVAC) systems, lighting, smart meters, Wi-Fi, people counting & occupancy sensors, access control systems, elevators, air quality sensors. These models are based on statistics, probabilistic models, machine learning, and deep learning. The understanding of how a building is used and how it performs under various conditions is used to devise and implement strategies to reduce operating costs, improve the user experience, and increase revenue. Our goal is to develop decision-making support systems and autonomous optimization routines, both at the building level and at the portfolio level (i.e., for clients that own several buildings). An example of decision-making support is a product that quantifies the possible yearly energy savings effectively obtainable by improving the thermal insulation of the building given knowledge of how the building is used. An example of autonomous optimization is the automated and dynamic optimization of the thermal comfort and the energy consumption of the building given current and predicted building occupancy.
Our data science technology stack is mostly Python-based (numpy/scipy, scikit-learn, Pandas, TensorFlow, Keras), but also includes Scala (Spark and Apache Beam). From a software engineering standpoint, our platform is built on top of Docker and Kubernetes, Kafka, Spark, Beam, Airflow, and related services from different cloud providers (Google Cloud Platform, Amazon Web Services, Microsoft Azure).
Members of the data science team are responsible for designing, implementing, and evaluating models and algorithms on a range of use cases, identifying new use cases from available data, as well as identifying promising and evaluating new approaches and technologies.
- To be enrolled in a master's or Ph.D. program in a quantitative discipline (e.g., statistics, operations research, bioinformatics, economics, computational biology, computer science, mathematics, physics, electrical engineering, industrial engineering) or equivalent educational program.
- Experience with statistical software (e.g., R, scientific Python, MATLAB)
- Willingness to develop innovative products, to perform research
- A solid background in mathematics and statistics
- Willingness to work in a team, review work done by others, and take constructive feedback
- Willingness to develop high quality models and software
- Written fluency in English
- Academic or industrial experience in a data analysis related field
- Applied experience with machine learning on large datasets
- Willingness to learn new techniques
- Written and spoken fluency in English
- Effective written and verbal communication skills
- Experience performing state-of-the-art research, writing high-impact academic publications
- Applied experience with Python for data analysis tasks
- Experience using relational databases and non-relational databases
What we offer
- An experienced data science team with a very strong research mentality yet focused on fast and agile execution to achieve business impact
- A data-centric product, where data scientists are at the core of the business
- A team of experienced and agile engineers responsible for making data available for use by various models and algorithms
- A competitive salary
- Fast-growing early-stage startup: team size is expected to double in the next 12 months
- A multi-cultural team that is passionate about technology, regular team outings
- Open communication, flat hierarchy, and fast execution
- Flexible working hours, allowance for partial telecommuting
- A comfortable office in the center of Paris (Strasbourg – Saint-Denis metro station)
Problem examplesThe data science team is regularly challenged to solve hard problems that may require the manipulation of uncertainty, the fusion of heterogeneous sensor data, and the modelling of complex interactions between behavioral or physical phenomena. Examples of problems that the team has worked on or is currently working on include:
- Develop statistical or machine learning models to estimate variables about human behavior in a building using sensor data.
- Use Wi-Fi data to estimate what is the probability to visit location A after visiting location B, while taking into account sampling bias inherent to the periodic and random nature of Wi-Fi measurements.
- Use depth sensing camera data to track people and their interactions with their environment within a complex space (e.g., kitchen).
- Use electrical energy consumption data of an elevator collected by a smart meter at a 1 Hz sampling rate to estimate the sequence of floors at which the elevator stopped and the number of people carried between every two consecutive stops.
- Develop models to predict human behavior and building performance under various circumstances.
- Predict the energy efficiency of a building after refurbishments given various occupancy scenarios.
- Predict what the activity level of a tenant will be in the next 3 months.
- Develop models to summarize human behavior and building performance and propose optimization strategies.
- Devise a behavioral model of shopping center users and find the most common behaviors within this model (e.g., weekend dining, weekday shopping, weekday lunching, late night entertainment).
- Identify optimization opportunities and propose new strategies for the heating and ventilation system of a building to optimize energy efficiency and thermal comfort with respect to its actual occupancy.
- Prototype solutions to help shape or meet specific business needs and goals.
- Collaborate with product management and engineering departments to understand business needs and devise possible solutions.
- Apply scientific rigor into the development of new solutions: prior art research, experimental methodology, hypotheses testing.
- Assess the quality of a solution with respect to how well it meets business needs and goals.
- Report and present experimental results and research findings clearly and efficiently, both internally and externally.
- Discuss research findings with other scientists and devise possible new directions and improvements to the devised solution.
- Propose possible new use cases to product management based on research findings.
- Write production-ready code for solutions that go past the prototyping phase.
- Collaborate with the product management department to define the scope of the proposed solution.
- Collaborate with the engineering department to devise possible production-ready solutions.
- Continuously improve the implemented solution by identifying issues, assessing its quality, and devising possible solutions.
- Work within a Scrum framework with other scientists and in collaboration with other departments.
To apply for this position, please send us your CV to firstname.lastname@example.org with the subject line "Data science intern".