Trine Siewartz Engelund Email & Phone Number

Copenhagen, Dk

København, Hovedstaden, Danmark

The AI unit at JP/Politikens Hus aims to unlock the potential of AI for newspapers such as Politiken, Jyllands-Posten, Ekstra Bladet, Finans, WatchMedia, etc.In my role, I am responsible for:• Leading AI projects as a Project Manager, including project planning and communication with stakeholders across the organization• Developing and researching Natural Language Processing (NLP) models to enhance the user experience of online news media• Planning and coordination of tasks for a student assistantMy expertise spans the entire ML model development lifecycle, encompassing data sampling, labeling, and preprocessing, as well as model training, evaluation, and monitoring. I have particular experience with fine-tuning transformers and LLMs as well as prompt engineering. In addition, I develop in-house Python packages and applications specifically designed for production use.Notable contributions include:• Built a framework and Python package with metrics for monitoring NLP models in production• Extending the development and deployment of NLP models originally trained for Ekstra Bladet to serve the broader media ecosystem within JP/Politikens Hus (ongoing project)• Fostering new initiatives in audio and voice AI technologies, including text-to-speech, speech-to-text, and voice cloning applications

København, Hovedstaden, Danmark

Ekstra Bladet is the largest online news media in Denmark and part of JP/Politikens Hus. I worked with developing Natural Language Processing models for online news articles at www.eb.dk. Due to the successes within our team, we moved on to the broader scope of working on AI initiatives for the entire media house of JP/Politikens Hus.Notable contributions include:• Fine-tuned a Danish transformer (BERT-like) for Sentiment Analysis and built a Flask application for production• Developed a Python package for classifying the topicality of news articles, incorporating prompt engineering for GPT-4 and fine-tuning of GPT-3.5• Created datasets for fine-tuning Named Entity Recognition, Topic Classification, and Sentiment AnalysisDuring my time in the team, we received the following awards for our work:• First place in the "Best Product Iteration" category at the INMA (International News Media Association) Global Media Award 2023 in New York.• The "Data and AI-Driven Transformation Project of the Year" award at the DAIR Awards 2023 in Stockholm

København, Hovedstaden, Danmark

As a student assistant on the Data Science team specializing in NLP, I undertook various responsibilities:• Conducting research and crafting annotation taxonomies and guidelines, along with actively participating in data annotation for finetuning transformers like BERT.• Handling diverse tasks such as NER, Topic Classification, Semantic Similarity, and Evergreens.• Developing a keen awareness of bias and trends in big data during the data annotation process.• Performing dataset preprocessing, including formats like CSV and JSON, using tools such as Pandas and Numpy.

København, Hovedstaden, Danmark

Student assistant at Alinea (part of Lindhardt & Ringhof and EGMONT). Alinea publishes teaching materials for children in primary school. I worked at the editorial office for children with special learning needs (e.g., dyslexia).My main tasks were to:• design drafts for books and websites• do proofreading and transcriptions of books and graphics• communicate with external graphic designers in India• migrate websites• do graphical work and design copy sheets (Adobe)• HTML editing in WordpressDuring my time at Alinea, I trained several other student assistants.

Lillerød

In my gap year, I worked as a Customer Advisor at Udbetaling Danmark ATP in the department dealing with the collection of debts.In this job, I advised citizens with debts and helped them set up installment plans. My job was to balance the company's interests with excellent customer service, as many citizens were often very emotional about their economic situation.During my time at ATP, I passed the internal education in customer handling and trained several other young advisers.I was employed by Randstad.

Frederiksberg

As a project employee my goal was to maximize the number of memberships by organizing events and motivating the volunteers by social media.

Master Of Science - Ms, It & Cognition, 11,23 / 12

I learned to combine IT and cognition to develop algorithms in Python mimicking human cognition, e.g., language, memory, learning, vision.

Guest Student, Data Science

Courses: • Text Analytics • Data Mining, Machine Learning, and Deep Learning • Artificial Intelligence and Machine Learning • Concepts.

Bachelor'S Programme, Linguistics

At Linguistics I learned to extract patterns and trends from large amounts of language data: grammatically, phonetically, phonologically.

Folk High School, Journalism And Media

High School, English, Mathematics And Social Studies

Primary School