Jonas Berhin
Data Scientist at constellr- Claim this Profile
Click to upgrade to our gold package
for the full feature experience.
Topline Score
Bio
0
/5.0 / Based on 0 ratingsFilter reviews by:
Experience
-
constellr
-
Germany
-
Aviation and Aerospace Component Manufacturing
-
1 - 100 Employee
-
Data Scientist
-
Aug 2022 - Present
Région de Bruxelles-Capitale, Belgique
-
-
-
Geckomatics
-
United Kingdom
-
IT Services and IT Consulting
-
ML Engineer/Geographer
-
Aug 2020 - May 2022
Antwerp, Flemish Region, Belgium Geckomatics builds mobile mapping systems powered by Machine Learning to identify objects and position them accurately. As expected from a position in startup, my role is polyvalent: ● Productionizing a structure-from-motion process to build city scale high resolution point cloud via: o Hundreds of orchestrated containerized services (Kubernetes, RabbitMQ & Github CD/CI) o Spark for non-parallelizable process like specific graph processes for example ● Dual Fisheye video flux… Show more Geckomatics builds mobile mapping systems powered by Machine Learning to identify objects and position them accurately. As expected from a position in startup, my role is polyvalent: ● Productionizing a structure-from-motion process to build city scale high resolution point cloud via: o Hundreds of orchestrated containerized services (Kubernetes, RabbitMQ & Github CD/CI) o Spark for non-parallelizable process like specific graph processes for example ● Dual Fisheye video flux fusion into equirectangular 360 and cube map frames (C++) ● Benchmarking and implementing meshing algorithms (CGAL/Open3d - C++ & python) ● Automating objects point cloud extraction based on image-based labels (scipy/spherical-geometry - python) ● Automating objects like Trees, Traffic Signs etc. characterization & smart measurements based on their individual point cloud (sklearn - python) ● Maintaining postgres/gis database and ensuring high data quality by automating checks etc. (networkx/geopandas – python) ● Leading projects deliveries and planning future research and development Show less
-
-
-
Deloitte
-
Business Consulting and Services
-
700 & Above Employee
-
Data Scientist
-
Jun 2018 - Jul 2020
Zaventem, Flandre, Belgique Data Scientist/Machine Learning Engineer at Deloitte Consulting : ● Predicting erroneous antibiotic treatment starts and stops in intensive care units (sklearn - python) ● Big Data Sales modelling for a consumer goods company (custom GLM - spark & python) ● Text extraction from ID pictures automated tool (CV - python) ● Predictive Maintenance at a consumer goods company production site (time series - R) ● Tax classification for tarrification internal tool development (NLP -… Show more Data Scientist/Machine Learning Engineer at Deloitte Consulting : ● Predicting erroneous antibiotic treatment starts and stops in intensive care units (sklearn - python) ● Big Data Sales modelling for a consumer goods company (custom GLM - spark & python) ● Text extraction from ID pictures automated tool (CV - python) ● Predictive Maintenance at a consumer goods company production site (time series - R) ● Tax classification for tarrification internal tool development (NLP - python) Show less
-
-
-
Ciel & Terre International
-
France
-
Services for Renewable Energy
-
1 - 100 Employee
-
Data Scientist/Remote Sensing Engineer - At Ciel & Terre International
-
Feb 2018 - May 2018
Lille Area, France Still in the context of lakes prospection for floating photovoltaics, once Ciel & Terre validated my PoC, I started to build several tools aiming to help them in the prospection. The tool is using Sentinel 2 API for smart image acquisition, several machine learning algorithms for the image processing and custom spatial crossover technics to associate contextual data from open-source API. The end product is coded in Python and R and integrated with ESRI (ArcGIS) products to leverage their… Show more Still in the context of lakes prospection for floating photovoltaics, once Ciel & Terre validated my PoC, I started to build several tools aiming to help them in the prospection. The tool is using Sentinel 2 API for smart image acquisition, several machine learning algorithms for the image processing and custom spatial crossover technics to associate contextual data from open-source API. The end product is coded in Python and R and integrated with ESRI (ArcGIS) products to leverage their existing UI and tools. The main challenges were developing from scratch to a finished product within a short period of time and without technical supervision (dealing with data management, API calls, data processing and integration with ArcGIS). But having such responsibilities was a great boost in term of creativity, experimentation, determination, etc.
-
-
Data Scientist/Remote Sensing Engineer - Developing the PoC
-
Nov 2017 - Jan 2018
France Ciel & Terre used to manually prospect for suitable lakes for floating photovoltaïc installation in Google-Earth/Google-Map. To grow their business Ciel & Terre were investigating new approaches to support their business developer. In that context, I developed a dashboard to showcase remote sensing and GIS technologies and how they could leverage them. Ciel & Terre were enthusiastic about it and they later offered me to work for them. The dashboard (R-Shiny), is still available… Show more Ciel & Terre used to manually prospect for suitable lakes for floating photovoltaïc installation in Google-Earth/Google-Map. To grow their business Ciel & Terre were investigating new approaches to support their business developer. In that context, I developed a dashboard to showcase remote sensing and GIS technologies and how they could leverage them. Ciel & Terre were enthusiastic about it and they later offered me to work for them. The dashboard (R-Shiny), is still available here, please take a look https://berhinj.shinyapps.io/APP_SAVE/ (optimization could be better, sorry it was the first time I was working with R-Shiny).
-
-
-
Les Jeunes Entreprises asbl (LJE)
-
Belgium
-
Education Administration Programs
-
1 - 100 Employee
-
Young Entrepreneur - RH
-
Sep 2009 - Jul 2010
Louvain-La-Neuve, Belgium During my junior high school year (16-17 years old), with few friends, we build a small business with the help of LJE asbl. We were promoting healthy and seasonal food habits in schools. Against all odds, it worked pretty well and by the end of the year, the action increased by over 30%. During this first approach to « the real world », I experienced: • How to develop an idea and find a direction • How to keep accounts • Dealing with diverging opinion
-
-
Education
-
Université catholique de Louvain
Master's degree, Geography -
Université catholique de Louvain
Bachelor of Science - BS, Geography