Publications GitHub Academic CV Resume Let's talk
Sharon Ibejih

Hello there 👋🏽, I'm Sharon Ibejih.

I am a machine learning engineer with research and industry experience. I analyze data, build and package machine learning models for production. I have also authored some interesting research papers. My research interests are low-resourced NLP, Automatic Speech Recognition and low-resourced data collection/generation.

Experiences

Sterling Bank Plc. - Lagos, Nigeria Data Scientist January 2023 - Present
Insidatics AI Inc. - Delaware, USA ML Software Engineer (Contract) February 2022 - October 2022
RectLabs Inc. - Lagos, Nigeria AI Engineer July 2022 - September 2022
Data Scientists Network (DSN) - Lagos, Nigeria Researcher & Solution Analyst October 2020 - April 2022

Selected Publications

NaijaNER: Comprehensive Named Entity Recognition for 5 Nigerian Languages

Most of the common applications of Named Entity Recognition (NER) is in English and other highly available languages. In this work, we present our findings on NER for 5 Nigerian Languages - Igbo, Yoruba, Hausa, Pidgin...

Read more

AFRIFASHION1600: A Contemporary African Fashion Dataset for Computer Vision

This work presents AFRIFASHION1600, an openly accessible contemporary African fashion image dataset containing 1600 samples labelled into 8 classes representing some African fashion styles.

Read more

EDUSTT: In-Domain Speech Recognition for Nigerian Accented Educational Contents in English

English ASR systems are trained on regular speech, therefore they may struggle to perform well on accented and domain-specific speech. Our experiment...

Read more

TCNSpeech: A Community-Curated Speech Corpus for Sermons

In this work we present TCNSpeech, a community-curated multispeaker sermon corpus for speech recognition tasks. It contains a total of 24 hours of English audio data recording, chunked and transcribed.

Read more

Selected Projects

illustration

Named Entity Recognition in PDFs

Named Entity Recognition (NER) is used to retrieve textual information of entities. Since NER is mostly helpful in large texts, this project is focused on scanning PDF docs to extract existing named entities.

Read more
illustration

Term Deposit Prediction

A term deposit is a cash investment held at a financial institution for an agreed rate over a fixed amount of time. With telephone being one of the most effective medium of communication with customers, it is crucial to identify the customers most likely to patronize term deposits beforehand so that they can be specifically targeted via call.

Read more
illustration

Nigerian Product Apps Analytics

NG App Analytics has made it easy for Product Owners to quickly check how their app is performing on PlayStore - in comparison to other similar Nigerian mobile apps. Owners can also analyse the reviews of their apps given by their customers on either Apple or Playstore.

Read more
illustration

Real Time Speech Recognition for Nigerian Church Sermons

How convenient would it be to have a digital transcriber of church sermons that understands the Nigerian accent? This work is an implemention of our research on TCNSpeech data of 24 hours audio duration.

Read more

Speaking Events

illustration

Datafest Africa'22: Who is a Researcher?

This topic targeted machine learning beginners who are curious about research.

Slides
illustration

AI Saturdays Lagos: Opportunities in Machine Learning from a Community Manager's Perspective

Slides
illustration

Computational Research and Open Science Community Mentorship Indaba

Represented She Code Africa to discuss how we manage mentorship programmes and projects.

Slides
speaking-event-banner

Machine Learning Experiment Tracking with MLflow

A hands-on lab on how to use MLflow Tracking for logging experiments.

GitHub Repo