Derek Nsoh Amuna

Specializing in the development, deployment, and security of machine learning systems. Building scalable solutions from the ground up with Deep RL, Generative AI, and cloud platforms.

Technical Skills

AI / ML & LLMs

PythonPyTorchScikit-LearnDeep RLRAGVector DBs

MLOps & Deploy

DockerFastAPIGitHub ActionsAWSGCPSQLPostgreSQL

Data & Analysis

RData MiningEconometricsFinancial AnalysisStatistical Modelling

Systems & Languages

GoCUDAC++GitLinux

Projects

A selection of things I've built — from an open-source Ghanaian speech-recognition model to production RAG systems and CUDA deep learning frameworks.

★ FEATURED PROJECT Open Source · Hugging Face

Griot Nano 1 — Multilingual ASR for Ghanaian Languages

A 153M-parameter Conformer CTC speech-recognition model trained on 44,000 audio-hours across Akan, Dagbani, Ewe, Ghanaian English, and Ga. Evaluated on 99,547 held-out recordings and released openly with inference scripts, KenLM decoding, and an ONNX export — so it can run fully offline on hardware people control.

5
Ghanaian languages
44,000
audio-hours trained
153M
parameters
13.66%
combined test CER
99,547
held-out test samples
ConformerCTCPyTorchASRAkanDagbaniEweGaGhanaian English

Experience

Data Science & AI – National Service Personnel

Sep 2024 – Sep 2025

  • Developed and deployed ML models using Scikit-Learn and FastAPI for live web applications
  • Managed the secure MLOps lifecycle using AWS and GCP for scalable deployments
  • Built automated data integrity pipelines to validate and clean sensitive datasets

Intern – Data Analysis

Aug 2023 – Sep 2023

  • Applied ML regression techniques for the annual budget and revenue growth plan
  • Conducted SKU-level contribution margin analysis and periodic sales reports

Director

Dec 2021 – Present

  • Developed ML regression models for sales prediction and strategic planning
  • Conduct ongoing data and financial analysis for analytics applications

Education & Certifications

University of Ghana

BA in Economics & Statistics

Graduated Feb 2025

Relevant: Statistical Computing, Econometrics, Macro/Micro-economic Analysis

Open Source

Live from GitHub