Salomon Kabongo

Expérience

• Lead Software Engineer

  State Farm — Innovation Group

  Feb 2022 – Lelu

  Applied AI research and engineering

  • Led applied research and development on synthetic-media detection, video understanding, and vision-language models to inform technical decisions.
  • Designed embedding-based pre-labeling pipelines to accelerate data annotation and model-development workflows.
  • Architected a document deduplication system combining exact hashing and visual-similarity search to identify near-duplicate content at scale.
  • Co-invented computer-vision systems for privacy and document analysis, including an issued U.S. patent and a pending published patent application.

• Board Member

  Masakhane Research Foundation

  2021 – May 2026

  Global

  Former board member

  • Helped shape the Masakhane AI Hub’s 2025–2029 strategy for language technology and digital public infrastructure serving African-language communities.
  • Supported governance and partnerships associated with approximately $9M in research funding and collaborations with Google.org, Lacuna Fund, UNESCO, the Gates Foundation, and IDRC.

• Research Assistant

  L3S / Leibniz Information Center for Science & Technology (TIB)

  Nov 2020 – Nov 2022

  Hannover, Germany

  • Defined and pursued a multi-year research agenda for converting empirical AI results in scientific papers into reliable, machine-actionable leaderboard records.
  • Developed and evaluated transformer and large-language-model pipelines across supervised, zero-shot, few-shot, and instruction-tuning settings.
  • Engineered the core Leaderboards feature for the Open Research Knowledge Graph, integrating extracted Task-Dataset-Metric-Score records into searchable knowledge-graph views.
  • Built datasets and evaluation workflows for scholarly information extraction, including context-selection studies, ablations, and exact- and partial-match analyses.

Dilonga

• PhD in Computer Science — AI / Natural Language Processing (LLMs)

  Leibniz Universität Hannover

  Nov 2020 – Nov 2025

  Hannover, Germany

• Master's in Machine Intelligence

  African Master's in Machine Intelligence (AMMI)

  Sponsored by Google and Facebook through AIMS

  Oct 2019 – Nov 2020

  Accra, Ghana

• Master's in Mathematical Sciences

  University of the Western Cape

  African Institute for Mathematical Sciences (AIMS South Africa)

  Aug 2018 – Jun 2019

  Cape Town, South Africa

• BSc (Honours) in Mathematics & Computer Science

  Université de Lubumbashi

  Oct 2014 – Jul 2017

  Lubumbashi, DRC

Bapublication ne ba brevet basungudibue

• Systems and Methods for Image Privacy and De-identification

  2026

  U.S. Patent

  U.S. Patent No. 12,613,996 · Issued Apr. 28, 2026 · Assignee: State Farm

  Issued U.S. patent

• INJONGO: A Multicultural Intent Detection and Slot-filling Dataset for 16 African Languages

  2025

  ACL

  Co-authored a culturally grounded benchmark for intent detection and slot filling in 16 African languages.

• IrokoBench: A New Benchmark for African Languages in the Age of Large Language Models

  2025

  NAACL

  Co-authored a human-translated benchmark for evaluating language-model reasoning across 17 African languages.

• Effective Context Selection in LLM-based Leaderboard Generation

  2024

  NLDB

  First-author study of how document representation and context selection affect extraction quality, reliability, and efficiency.

• Systems and Methods for Advanced Duplicate Image Search and Analysis

  2024

  U.S. Published Patent Application

  App. 18/652,500 · Publication No. US20240411724A1 · Assignee: State Farm

  Pending published application

• Zero-shot Entailment of Leaderboards for Empirical AI Research

  2023

  ACM/IEEE JCDL

  First-author evaluation of whether supervised extraction models generalize to unseen leaderboard labels.

• Bibletts & LiSTra: African Speech Corpora

  2022

  Interspeech, NeurIPS Workshops

  High-fidelity multilingual speech corpus; first English-to-Lingala speech translation baseline

• Automated Mining of Leaderboards for Empirical AI Research

  2021

  ICADL · International Journal on Digital Libraries

  First-author research on extracting and representing empirical AI results from scientific text.

  ICADL 2021 Best Paper Award

• Participatory Research for Low-Resourced Machine Translation

  2020

  EMNLP Findings

  Community-led research on methods and infrastructure for African-language machine translation.

Ba compétence technique

Ba distinction

• 2026

  U.S. Patent No. 12,613,996 issued

  Systems and Methods for Image Privacy and De-identification · Assignee: State Farm

• 2021

  ICADL Best Paper Award

  International Conference on Asian Digital Libraries

• 2020

  DLRL Summer School

  CIFAR / Mila, Montreal

• 2020

  Google Hash Code — Ranked 1747/10724

  Google

• 2019

  ACM Future of Computing Academy (FCA) Member

  Association for Computing Machinery — 36 selected globally