Abdallah Bashir

Applied Scientist II

Goldman Sachs

Biography

Hi! I am Abdallah. Welcome to my homepage. I am an Applied Scientist at the NLP Modleing team in Goldman Sachs. I have obtained my MSc degree in Computer Science from Saarland University. My thesis focuses on leveraging self-supervised learning in domain specific languge models. My research interests include solving real world NLP problems. I can be classified with high confidence as a coffeeholic Ambievert who loves to talk about comedy, travel (below images are my own), politics, and philosophy.

Download my resumé.

Interests

Machine Learning/Deep Learning.
Natural language processing (NLP).
Medical/ Financial NLP.
Large Langauge Models LLMs (pertaining, fine-tuning, synthesizing data).
Search (retrieval/ reranking).

Education

M.s in Computer Science, 2021
Saarland University
B.S. in Electrical and Electronic Engineering, Software Engineering, 2018
University of Khartoum

Experience

Applied Scientist II

Goldman Sachs

Aug 2022 – Present London

Spearheaded the development and integration of advanced machine learning and deep learning models, with a focus on Natural Language Processing (NLP) and Large Language Models (LLMs). Trained and fine-tuned different classes of models for NLP usecases.
Innovated in areas such as retrieval and reranking, retrieval augmented generation (RAG), fine-tuning LLMs, text classification, named entity recognition, semantic parsing, and synthesizing training data to be used in various model training exercises. Significantly advancing the capabilities of LLMs in interpreting and processing human language for enhanced customer experiences.
Proficiently utilize Python, PyTorch, and distributed computing frameworks for efficient model training on GPU clusters. Design and maintain deployment pipelines with continuous integration, version control, and monitoring.
Comfortable diving into ambiguous problems by conducting extensive literature reviews to understand the state of the art at the time. Drive and implement solutions that address complex challenges, transforming research findings into production-ready code.
Achievements - Co-developed two patent-pending frameworks and implemented ML solutions that realized a cost reduction of 10M USD, demonstrating a direct impact on operational efficiency and innovation.

Applied Scientist Intern

Amazon

Apr 2022 – Jul 2022 Seattle, WA

Using Transformers-based models, Implemented an end-to-end unsupervised framework to identify sensitive trending topics in Alexa traffic.
Achievements - the framework will increase the robustness of online sensitive detection models in Alexa by refining the quality of labeling data.

Research Intern

Allen AI

Jan 2022 – Mar 2022 Seattle, WA

Conduct research in Natural Language Proceeding Data Augmentation using Language models using LSTMs and Transformers
Implemented pipelines that leverage models like Albert and BART to extrapolate datasets with structured entities.

Applied Scientist Intern

Goldman Sachs

Apr 2021 – Aug 2021 Frankfurt, DE

Curate, parse, preprocess data related to git projects in the organization.
Implementing Deep learning models with Attention to identify high risk changes in Git Software products.
Achievements - Implemented a Machine Learning based tool to help code reviewers to detect high risk changes in software product

Research Assistant

Max Planck Institute for Informatics

Nov 2019 – Mar 2021 Saarbrucken

Conducted groundbreaking research focused on integrating contextual understanding into neural language models through contrastive learning approaches.
Implemented models in diverse applications including healthcare, achieving significant reductions in model size and computation time while maintaining competitive accuracies.
Developed novel model architectures utilizing Knowledge Graphs to enhance model performance in NER and QA tasks across specialized domains.

Intern

Dialog Systems and Machine Learning - HHU

Aug 2019 – Sep 2019 Düsseldorf

Ported Pydial 2.0 to Pydial 3.0 (an open-source statistical dialogue systems tool).
Achievements - Wrote and reviewed code for Experiments, unit tests, and evaluation of the porting process. ported the library with 0% errors.

Research Assistant

German Research Center for Artificial Intelligence ( DFKI )

May 2019 – Jul 2019 Saarbrucken

Built an API to preprocess text data and get it in shape for modeling.
Achievements - trained Classifier by fine-tuning BERT to classify post machine translation errors.

Co-Founder, Software Engineer

EchoNine IT Solutions

Sep 2015 – Oct 2016 Khartoum

Gathered and defined customer requirements to develop clear specifications for creating well-organized project plans.
Designed, built, and monitored web/mobile applications and sites for continuous improvement.
Achievements - building a product in cooperation with Zain Sudan, the top telecommunication company in the country.