Moussa Koulako Bala Doumbouya

I (AKA Moussa Doumbouya): [Google Scholar] [LinkedIn] [Twitter] am a fifth-year Computer Science Ph.D. student at Stanford University, advised by Christopher Manning and Dan Jurafsky. My research focuses on Artificial Intelligence (AI), Natural Language Processing (NLP), and Computational Pedagogy (use of algorithms to enhance human learning). In particular, I explore how machine learning can augment pedagogy and how insights from psychology can guide the development of human-interpretable machine learning algorithms.

I am also committed to advancing STEM education in West Africa and the inclusion of underserved languages in NLP. My recent projects span linguistic resource curation, machine translation, speech recognition, computer science education, AI safety, computational pedagogy, and psychologically plausible neural networks.

Prior to Stanford, I worked as a software engineer and a research engineer at Libera, Agero, EBSCO, VideoIQ/Avigilon/Motorola, and Apple. My work ranged from enterprise software development and large-scale computer vision for video analytics to neural network interpretability research for low-power devices and autonomous navigation systems.

Recent Projects

• Tversky Neural Networks: Psychologically Plausible Deep Learning with Differentiable Tversky Similarity (Under peer review at NeurIPS 2025)
• h4rm3l: A Dynamic Benchmark of Composable Jailbreak Attacks for LLM Safety Assessment. (ICLR 2025)
• Handwritten Code Recognition for Pen-and-Paper CS Education. (L@S 2024)
• Machine Translation for ߒߞߏ (NKo): Tools, Corpora, and Baseline Results. (WMT 2023)
• Using Radio Archives for Speech Recognition in Maninkakan, Pular and Susu. (AAAI 2021)

Teaching

• 2019-2020: GNCode Coding Camps
• 2020-2021: Code in Place
• 2024: Stanford CS224N Spring 2024

Publications & Links

• Academic Publications & Patents: [Google Scholar]
• Curriculum Vitae: [CV]
• Social Media: [LinkedIn] [Twitter]

Blog

• 2025-05-02: Selfie, Selfies, Selvie, Selvies