Chima Emmanuel/About & Bio

About Me — Chima Emmanuel

Founder, Modular · ML researcher & engineer · Nigeria

I'm Chima Emmanuel Akuegbo, a machine learning researcher and engineer, and the founder of Modular, an AI and data infrastructure company building open models, datasets and pipelines, with a biotech division.

At Modular I lead the research: code-switching speech models for Nigerian languages, multilingual embedding and vision–language models, a small time-series foundation model, and protein stability prediction. We release our work openly on Hugging Face, with model cards that state what works, what does not, and what each model was allowed to learn from.

Alongside Modular, I'm Lead AI Engineer at Phace LLC (AnomalyForex) and co-founder of Llava, an AI operating system for lawyers, and of Forecast, a Nigerian prediction market. I study Software Engineering at the Federal University of Technology Owerri, and I contribute to open-source ML infrastructure, most recently PyTorch's ExecuTorch.

I write about building serious R&D where the infrastructure is thin. Start with the essay and the short book on the Publications page.

Research Focus

Speech for how Nigerians actually talk

Text-to-speech, speech recognition and voice agents for English, Pidgin, Igbo, Yoruba and Hausa, built for code-switching inside a single sentence.

Multilingual & multimodal representation

Embedding and vision–language models that put Nigerian languages and images in one shared space for search, retrieval and grounding.

Small foundation models

Compact models trained from scratch for the GPU-poor, such as a 2.4M-parameter time-series forecaster that runs on a phone-class device.

ML for the life sciences

Structure-conditioned protein stability prediction and other tools for Modular’s biotech work.

Edge & on-device ML

Getting models onto small hardware, including open-source contributions to PyTorch’s ExecuTorch runtime for microcontrollers.

Research under constraints

Licence-clean data, staged experiments on rented compute, and honest evaluation where the usual infrastructure is missing.