Shippeo, the European leader in supply chain visibility, gives shippers, carriers, and end-customers instant access to predictive and real-time information of every delivery.
Shippeo’s machine-learning, proprietary algorithm dynamically calculates ETAs allowing shippers to quickly anticipate problems, proactively alert end-customers, and efficiently manage exceptions. Shippeo helps market-leading companies leverage transportation to deliver exceptional customer service and achieve operational excellence.
Founded in 2014, Shippeo tracks more than 3 million loads per year throughout Europe and connects to carriers in more than 20 countries. Shippeo’s more than 160 employees have 22 different nationalities and speak 27 languages!
We are looking for an Intern in Data Science to join our Analyze & Predict tribe.
The Analyze & Predict tribe is responsible for leveraging the large amount of data that
Shippeo has been acquiring over the course of running the platform and rolling it out to
multiple shippers and carriers, to get insights from it.
One of the main features the team builds and improves is Shippeo’s proprietary Machine
Learning algorithm that predicts Estimated Times of Arrival (ETA) of trucks, which is an
extremely difficult exercise due to all the uncertainties in road transportation (traffic, weatherconditions, driving regulations, time spent on on loading or delivery site, milk-runs thatcontribute to error propagation, etc.). We are constantly looking for new ways to make the
ETA prediction as accurate and reliable as possible, to help our users anticipate delays.
In Shippeo platform, ETA is mainly used to answer the following customer needs :
1. When exactly will my truck arrive at loading/delivery?
2. Is there a risk of delay on my loading/delivery?
To meet these needs, we need to accurately predict the arrival time (ETA) to the minute and
the risk of delay. More formally, we need to develop a regression machine learning model
(M1) that accurately predicts the ETA (e.g., arrival on 14/7 at 14:15) to answer the first
question, while for the second question, we need a classifier (M2) that predicts the
probability that the truck will be late. However, the independent implementation of two different models can lead to inconsistency in our predictions, as we may eventually predict a delay using M2 while the predicted ETA shows no delay (using model M1).
This internship will consist in addressing the above mentioned problematic by exploring
multiple strategies (e.g. Models stacking), more specifically by developing a Multi-task
learning framework in which multiple objectives (ETA prediction & Delay prediction) are
trained within the same model at the same time (Zhang et. al 2021).
Interview process :
1. Preliminary call
2. Case study preparation
3. Case study presentation and final interview
Le secteur de la cybersécurité souffre d'une pénurie de talents face aux menaces actuelles. La formation dans le domaine reste la réponse...
User Experience Designer, expert en Millennial, YouTuber ou encore chauffeur de VTC ; les dix ans de technologie que nous venons de vivre...
Nous vivons dans un monde hyper connecté, où de plus en plus d'entreprises fournissent des ordinateurs portables et des smartphones dans...
Si les algorithmes régissent notre vie virtuelle, ils peuvent également être de véritables atouts pour gagner en performance et en...
Inscrivez-vous pour accéder à l'annonce
Data Scientist - InternshipEn cliquant ci-dessus sur « S'inscrire » vous accepter nos conditions générales et notre politique de confidentialité
1er Site de recherche d'emplois dédié aux professionnels du marketing de la communication et du digital, Jobibou.com a pour objectif de vous offrir le meilleur outil de recherche pour vous accompagner, au mieux, dans votre démarche de recherche d'emploi
Inscrivez-vous pour accéder à l'annonce
Data Scientist - InternshipMerci de bien vouloir remplir le formulaire pour profiter de l’ensemble des fonctionnalités Jobibou !
Déjà inscrit ? Se connecter