Data Scientist (m/w/d)

Kunde Ein Startup, das einen Autounfall-Assistenten entwickelt
Kategorie Data Science
Veröffentlicht
Uplink-Provision 10% für 12 Monate Mehr Infos
Job-Typ Direkt Mehr Infos

The client is looking for a Data Scientist, who wil be responsible for the development of action and cost prediction models in car insurance assisted by computer vision. Computer vision technology determines the damaged part of the car and classifies the severity/type of damage, and provides a detailed item-level invoice of the parts that would need replacement, and the additional labour that would be required to fully repair the vehicle. You will work together with the two CEOs.

Your tasks:
- Cleaning the data: in their dataset they have blended images of all vehicles involved in a collision, not properly tagged, and they need to filter the output of processing the images of the set with the relevant invoice
- Determining action prediction correlations: What type of damage in which part of the car probabilistically results in which repair task?
- Determining cost prediction correlations: What correlations can be inferred between the various parameters (part of the car, part type, type of damage, car model, etc) and the labour and parts cost?
- Developing such prediction models, and assessing their accuracy

Must have skills
Amazon Web Services, Python

Startdatum
Frühestens: sofort
Spätestens: 26. Jul 2021
Start should be asap.

Laufzeit
2-4 Monate
The project can be extended.

Auslastung
Vollzeit bevorzugt
You should be available at least 4 days/week with the possibility to downgrade to part time after some achieved milestones.

Remote
Teilweise möglich
Mostly remote, 1-2 onsite sessions in Berlin would be great.

Erforderliche Sprachkenntnisse
English is totally fine.

Budget
max. 90€ pro Stunde
inkl. Uplink-Gebühr

Einsatzorte
Berlin

Dieser Job ist bereits geschlossen und Bewerbungen sind daher nicht mehr möglich.
Um in Zukunft Jobs wie diesen nicht mehr zu verpassen, melde dich jetzt bei Uplink an!

Jetzt Mitglied werden