Machine Learning Research for Sustainable Development Goals (NLP Track)

Contract

This is a UNV contract. More about UNV contracts.

An SDG Integration Team located with UNDP’s Global Policy Network (GPN) offers a menu of services emphasizing direct short- to medium-term engagements to respond rapidly to requests from country offices for support on national implementation and monitoring of integrated policy solutions, qualitative and evidence-driven analysis for accelerated progress, and knowledge sharing and upscaling of innovative approaches to sustainable development. The team’s work emphasizes the application of evidence- driven data and analytics for SDG implementation and reporting. In this regard, advances in digital technology are creating data at unprecedented levels of detail and speed, turning the stories of people’s lives into numbers every minute of every day, across the globe. An important focus of the integration work is to complement traditional data (e.g., national statistics,) with new and alternative sources including digital ‘breadcrumbs,’ satellite data, social media to identify emerging trends and gain new perspectives on issues in development.

The team is looking for an experienced machine learning (ML) volunteer to support the team's research in the area of natural language processing (NLP) for Sustainable Development Goals (SDGs). The purpose of this assignment is to design, develop and test ML solutions for NLP problems in the area of SDGs. This is a research-intensive assignment that requires performing a range of tasks from literature review and prototyping models to model evaluation and writing up results. The research is expected to produce an industry-grade open-source model. Note that this is a highly selective opportunity. Your application must include a short motivation statement that clearly describes how your education and experience align with the task. You should include links to your GitHub, publications or online portfolio, if applicable.

• Master's degree in Computer Science, Data Science, Artificial Intelligence, Machine Learning, or Natural Language Processing is a prerequisite.
• PhD degree is an advantage.
• Strong mathematical background is required.
• Fluency in Python and relevant deep learning frameworks is required, e.g., PyTorch (preferred), TensorFlow, JAX.
• Proven knowledge of state-of-the-art deep learning and NLP techniques is required, including but not limited to BPE tokenisation, contextual representations, (self-)attention, transformers, encoder-decoder architecture zero- and few-shot learning.
• Documented research experience or publications in the area are a strong asset.
• Proficiency in English is required. Working knowledge of any other official UN language is an asset.