Mardiyyah Oduwole is an ML engineer and data systems architect known for designing and deploying machine learning systems from the ground up—from data pipelines and vector databases to production inference and monitoring.
Available for consulting on RAG systems, model compression, data infrastructure, and production ML architecture.
Designing production ML systems
Data infrastructure at scale
Model compression & efficiency
A career shaped across ML engineering, data systems, and production deployment.
Machine Learning Engineer
01/2026 – Present
Designed and deployed PeppaSync, a production ML system serving 50+ Shopify merchants with 99.9% uptime. Built revenue shield and growth engine systems.
Accelerator Engineer
03/2025 – 06/2025
Cut information retrieval time from 2 hours to 5 minutes (24x improvement) through production RAG system serving 100+ staff.
Data Infrastructure Engineer
09/2024 – 12/2024
Reduced data latency from 24 hours to 2 hours. Designed cloud-native ELT pipeline processing 10,000+ records daily.
Research Engineer
06/2023 – 05/2025
Achieved 64% model compression and 52% inference speed improvement on AfriBERTa using INT8 quantization.
ML Engineer
06/2022 – 06/2023
Reduced customer response times by 70% through conversational AI implementation.
Leadership areas shaped by real production environments.
Multilingual mental health chatbot in English, Yoruba, Igbo, and Hausa.
AI-powered research assistant for literature reviews and paper analysis.
Production ML system for Shopify merchants combining revenue protection and growth.
Cloud-native ELT pipeline for real estate data processing.
Academic and scholarly work in ML and African language NLP.
IJCNLP-AACL 2023 (Best Paper Award)
Large-scale multilingual news classification dataset for 16 African languages.
AfricaNLP @ ICLR 2024
64% model compression and 52% inference speed improvement on AfriBERTa.
EMNLP 2023
Multilingual sentiment analysis dataset and models for 11 African languages.
ACL 2021
Large-scale collaborative project for low-resource African language translation.
WMT 2023
Analysis of word embeddings for African languages and improvement recommendations.
Writing and reflections on ML systems, data engineering, and production deployment.
Coming soon. I'm working on sharing insights from building production ML systems.
Reach out for consulting, advisory work, or a thoughtful conversation.
GitHub
github.com/hemhemoh ↗