image

Nellie Duhs

Nellie is a software engineer with a positive mindset and passion for mathematics and logic. She is curious and very optimistic in her approach and has a profound love for problem-solving in general. She enjoys working with in a team with a shared goal and always makes sure that everyones feels included.

As a part of the inhouse team, Nellie is mainly working with machine learning, image analysis and as a backend developer.

Software Development Machine Learning Image Analysis Data Analysis

Programming languages

Languages Experience Rating Last used
Python 6 years 5 2025
SQL 5 years 5 2025
Typescript 2 years 4 2025
Java 3 years 4 2019
React 3 years 4 2024
React Native 1 years 3 2020
MATLAB 3 years 3 2018
C 2 years 3 2025
C++ 1 years 3 2021

Development tools

Tool Experience Rating Last used
Git 6 years 5 2024
BalenaCloud 4 years 5 2024
Rest API 4 years 5 2024
Arduino 1 years 3 2020
Gerrit 2 years 3 2019
Bitbake/Yocto 1 years 3 2019
Visual Studio Code 4 years 4 2024

Knowledge & technologies

Tool Experience Rating Last used
OpenCV 2 years 5 2022
Data Visualization 6 years 5 2024
Data Analysis 3 years 4 2024
NextJS 1 years 4 2024
Machine Learning 5 years 4 2024
Database Design/Management & Security 3 years 4 2024
Docker 4 years 4 2024
Algorithms 5 years 4 2023
Artificial Intelligence 5 years 4 2023
Keras 4 years 4 2023
Tensorflow 3 years 3 2023

Languages

Language Skill level
Swedish Native speaker
English Near native / fluent

Education

  • Master of Science, Engineering Mathematics (2014 - 2020) Lund University
  1. Has some knowledge of the technology/product.
  2. Has previously used the technology/product but may require a brief introduction or course.
  3. Has used this technology/product in projects before and can get going on his/her own.
  4. Feels very confident with the technology/product.
  5. Is very experienced with the technology/product and can support or educate others in this area.

Projects & Employments

Current & previous projects

Software developer at Volvo Cars, Lund

Working in the System Management team developing a tool to help analyze the performance on the Core Computer of Volvo's fully electric cars. Extracting large amount of data about the performance and resources of every application during test drives and visualizing the data.

Technologies: Python, SQL, Docker, JavaScript, Ninja, CSS, Gerrit

Backend Developer at Dirma, Helsingborg

Dirma is an innovative online platform simplifying the way businesses can sponsor associations through targeted marketing. Users create a profile and subsequently earn money to their favourite associations by receiving targeted email campaigns.

Technologies: Typescript, React, NextJS, Clerk, Prisma, Jest, Docker, Github CI/CD, Scaleway

Data Scientist at AOB Travel, Helsingborg

AOB Travel is a travel agent based in the south of Sweden. They came to us since they collect great amounts of data on search behaviours and booking information, stored in Kibana, but no way to visualise and make sense of it. For this project I started with exploratory data analysis, taking inventory and building an understanding of the data. I then developed advanced search patterns and dashboards in Kibana that allowed the customer to display the data in a way that enabled them to make decisions based on it.

Technologies: Kibana, Python, Elasticsearch

Machine Learning Engineer at Helsingborg Stad, Helsingborg

Development of a smart toilet seat (Helleringen) that can alert caregivers of patients with dementia about bowel movement issues, without encroaching on privacy. This is achieved by utilising radar for data capture and a trained neural network to determine whether it's "poop or not poop". Project scope included a hardware prototype for gathering the training data, a backend with a simple web app for gathering ground truth, and designing and training the model.

Technologies: Python, Machine Learning, Tensorflow, Keras, Radar, C, JavaScript, Linux, Docker, Balena

Software Developer at Agroväst, Helsingborg

This project which aims to help farmers by accurately determining the growth potential of individual seeds put inside the machine. To achieve this, a physical device, the Germination Box, was created, along with a sophisticated backend and image analysis solution. Seeds are placed in a grid and the optimal growth conditions are then met and held for about 48-72 hours. The device then takes continuous photos and sends them for analysis.

Technologies: Python, OpenCV, React, React Native, JavaScript, Docker, Linux, Raspberry Pi, DigitalOcean Cloud

Software Developer at EUROP, Helsingborg

EUROP is a project that aims to develop a computer vision based machine learning system to automate the classification of meat to European standards. Our team is responsible for the development of hardware and related software that ties together different inputs, such as images from cameras and different sensor inputs. I have been working on extracting information from physical labels by the means of OCR, object detection (Yolov3) and image analysis (OpenCV, scikit-image), developed the backend solution on digital ocean and developed a QR-scanner to retrieve information from physical labels.

Technologies: Python, Keras, Tensorflow, PyTesseract, Yolov3, OpenCV, scikit-image, pyenv, Docker

Machine Learning Engineer at Smart Agritech, Helsingborg

Pig Scale - a solution for determining the weight of pigs using image analysis. My task was to improve the detection and outlining of pigs that were standing in a desired position. This I did by creating a dataset with the wanted outlines and training a modell that performs instance segmentation. The modell was based on an open source implementation of Mask RCNN, which was written i Python, built with Keras and Tensorflow. This was optimised and fine tuned to generate both detection and segmentation with high precision. I also developed tools to visualize the results in a nice and smooth way; both to evaluate models and to be able to compare them with previous solutions.

Technologies: Python, Keras, Tensorflow, Tensorboard, VoTT, OpenCV, Matplotlib, IBM Cloud

Masters Thesis Student at Axis Communications AB, Lund

For my masters thesis I worked with machine learning and natural language understanding to develope a dialogue system for embedded devices. We collected and processed a dataset, designed and trained a recurrent neural network using Tensorflow and Keras and ran predictions with the resulting model on an intercom product using Tensorflow Light. We also implemented preprocessing of speech input data and the actual dialogue system that generated correct responses and/or actions based on the predictions from the model. This we did using C++ and open source libraries for speech recognition and generation.

Technologies: Python, Keras, Tensorflow, Tensorflow Light, Yocto/Bitbake, C++, Gerrit, Git

Summer Intern at Axis Communications AB, Lund

During the summers me and two other interns developed an adaptive volume application that adjusted the volume in a zone of network speakers based on the ambient noise in the room. We developed software to collect the volumes from every speaker and algorithms for the volume adjustment in C, used MQTT for communication between the speakers and used yocto/bitbake to deploy it to the speakers.

Technologies: C, MQTT, Yocto/Bitbake, Gerrit

Current & previous employments

EC Solutions AB, Helsingborg

Consultant - Developer at the in-house department working mainly with image analysis, machine learning and as backend developer.

Axis Communications AB, Lund

Summer intern and student worker working with software development and quality assurance for embedded devices.