Hi there...

My name is Ahmed Salem Mohamed Elhady...or just Salem.

I am a PhD Research Student at the University of the Basque Country (UPV/EHU) under the supervision of Mikel Artetxe and Eneko Agirre.

I am working on the analysis and improvement of multi-lingual capabilities in Large Language Models, with a focus on the Intersection of Reasoning and Multilingualism.

Before embarking on my PhD, I worked as an Applied Data Scientist at Microsoft Bing’s news and content moderation and understanding (now part of web and windows experience). I also worked as an Applied Data Scientist at Implicit.ai (formerly Agolo), a Microsoft-backed startup, where I worked on multiple projects building knowledge graphs and summarization for elite clients such as Franklin Templeton, Acuity, and The Associated Press. All the projects are part of Microsoft’s Azure services now.

News and Highlights

  • May 16th, 2025: 2/2 Papers Accepted to ACL 2025 - main track

    1. Emergent Abilities of Large Language Models under Continued Pretraining for Language Adaptation

    2. Wicked: A Simple Method to Make Multiple Choice Benchmarks More Challenging

  • February 26th, 2025

    New Preprint: "Wicked: A Simple Method to Make Multiple Choice Benchmarks More Challenging"

  • News Article about the group's main project

  • April 15th, 2024: our paper on “guided bart” is accepted to NAACL’2024.

    We presented it in oral presentation on June 18th, in Mexico City, Mexico.

  • November 5th, 2023: Our approach, guided bart, ranks 1st on the Cochrane subset of Allen AI’s MSLR shared task

What I do

  • design icon

    Freelance Projects

    I am top rated freelancer on Upwork with 94% job success rate.

  • Web development icon

    Open Source Contribution

    I am a contributor to a soon-to-announce open source project that aims to improve the document analysis in Arabic.

  • quote

    Quips, Quirks, and Jots

    I ponder and write down jots about Religion (Islam), Work, Research, and Others.

  • camera icon

    Street Photography

    I am a street photographer and I love to capture the beauty of the world around me. Checkout my instagram account for photography.

Daniel lewis
quote icon

Daniel lewis

Richard was hired to create a corporate identity. We were very pleased with the work done. She has a lot of experience and is very concerned about the needs of client. Lorem ipsum dolor sit amet, ullamcous cididt consectetur adipiscing elit, seds do et eiusmod tempor incididunt ut laborels dolore magnarels alia.

Resume

Experience

  1. HiTZ Center, University of Basque Country

    PhD Student - Research Assistant

    MAR 2024 — FEB 2028

    Working on analysis and evaluation of multilingual capabilities in large langyage models.

  2. Microsoft

    NLP - Applied Data Scientist II

    AUG 2023 – MAR 2024

    Working on Microsoft News’s text feed content understanding and moderation; Using LLMs and Transformer-based models for QA, retrieval, and misinformation identification.

  3. Implicit (formerly Agolo)

    NLP - Applied Data Scientist II

    JUL 2020 – JUL 2023

    - Worked on Conditional Text Generation for Table-to-Text generation using Parameter Efficient Fine-tuning

    - Worked on Knowledge Graphs and Multilingual Relation Extraction

    - Worked on Query/Ontology Based Summarization, Knowledge Graphs, and Query Understanding/Intent Classification.

    - Built Production Ready Deep Learning Models with Hessian-aware Quantization, Distillation, and Self-Supervised Learning

Education

  1. University of the Basque Country (UPV/EHU)

    2024 — 2028 PhD

  2. Zewail City of Science and Technology

    Feb 2022 – June 2024 Master of Science

    Artificial Intelligence Program, Department of Mathematics and Computer Information

    GPA: 4.0/4.0

    Thesis Topic: Improving Factual Accuracy in Multi-document Summarization of Clinical Documents

  3. Cairo University, Faculty of Engineering

    SEP 2015 – JUN 2020 Bachelor of Science

    GPA: 3.95/4.0, Rank in Class: 2nd

    Graduation Project: Grade: A∗, Topic: Improving Coherence, Readability, and Saliency in extractive summarisation of news articles.

Languages

  • Arabic
    100%
  • English
    90%
  • Spanish
    60%

Publications

  • Emergent Abilities of Large Language Models under Continued Pretraining for Language Adaptation

    Authors: Ahmed Elhady, Eneko Agirre, and Mikel Artetxe

    Venue:

    ACL 2025, Main

    [Paper] [X-thread]

  • WiCkeD: A Simple Method to Make Multiple Choice Benchmarks More Challenging for Language Adaptation

    Authors: Ahmed Elhady, Eneko Agirre, and Mikel Artetxe

    Venue:

    ACL 2025, Main

    [Code]" [Paper] [X-thread]

  • Improving Factuality in Clinical Abstractive Multi-Document Summarization through Guided Continued Pre-training

    Authors: Ahmed Elhady, Khaled Elsayed, Eneko Agirre, and Mikel Artetxe

    Venue:

    NAACL 2024, Main

    [Paper] [X-thread]

Contact