Professional Summary

Data scientist and founder building DNClimate, a venture helping African heavy industry exporters navigate the EU Carbon Border Adjustment Mechanism from 2026. Currently pursuing an MSc in Environmental Engineering at the University of Southern Denmark, with data science tenure at the Ghana Statistical Service. Interested in the intersection of trade data, industrial energy systems, and climate regulation."

Current ventures
Founder, DNClimate
  • •Published the Ghana CBAM Exposure Brief mapping the country's most exposed heavy-industry exporters.
  • •Launched the venture landing page at dnclimate.com⁠
    • 07/2026 – Present
  • •Joined the SDU Startup Station in September 2026.
    • Founder, Dataverse Ghana
  • •Established Dataverse Ghana as a foundational data consulting entity, defining its core business strategy, future service architectures, and long-term operational vision.
  • •Spearheaded the launch of The Dataverse Blog as the organization's initial educational arm, actively building brand authority and fostering a community around data literacy prior to launching B2B commercial operations.
    • 05/2025 – PresentAccra, Ghana
  • •Directed overall brand positioning, digital presence, and outreach strategy to position the entity as an emerging, trusted thought leader within the local data tech ecosystem.
    • EXPERIENCE
    Data Scientist, Ghana Statistical Service

  • •Analyze large datasets from diverse sources, including the 2021 Population and Housing Census.
    • 11/2023 – 08/2026Accra, Ghana
  • •Clean and process survey responses including the 2024 Ghana Marine Fishing Pilot Survey using Stata, generating accurate summary tables
  • •Develop data visualizations for survey reports including the Ghana Industrial Travellers Survey, 2024 Integrated Business Establishment Survey Phase I, and Domestic and Outbound Tourism Survey, that highlight trends, anomalies, and key insights for stakeholders
  • •Respond to client-specific data requests and deliver tailored analyses to meet project needs

    • EDUCATION
    Msc. Environmental Engineering, University of Southern Denmark
  • •Concentrations: Environmental Systems Analysis, Life Cycle Assessment (LCA), and Environmentally Efficient Technologies
  • •Related Coursework: System Analysis - Life Cycle Assessment, Material Flow Analysis, Energy System Analysis, Techno-economic Assessment of Process Technologies, and Geographic Information Systems (GIS)
    • 09/2026 – PresentOdense, Denmark
    BSc. in Actuarial Science, University of Professional Studies
  • •Concentrations: Risk management using probability and statistical methods, data analysis with R
  • •Related Coursework: Probability, Mathematics, Statistics, Finance, Economics, Financial economics, and Computer programming
    • 08/2019 – 08/2023Accra, Ghana
    1 / 3
    SKILLS
    Programming and Data Analysis — Proficient

    Python, R, PostgreSQL, Stata

    Data visualization & Dashboards — Expert

    Tableau, Looker Studio, Streamlit, Matplotlib, Seaborn, ggplot

    Tools & Platforms — Proficient

    Jupyter Notebook, RStudio, RMarkdown, Git & GitHub, Microsoft Excel, Google Analytics, Canva, Cloud Deployment (Render, AWS/GCP basics)

    PROJECTS
  • •Developed a micro-nowcasting classification engine to predict labor market participation and employment status (ICLS standards) using socio-demographic features.
  • •Trained and evaluated CatBoost models on ~595,000 records from the AHIES dataset, engineering a strict temporal holdout validation strategy to ensure realistic forward-prediction capabilities.
    • 03/2026
  • •Achieved a 0.63 Macro F1 score on a 3-class target with zero temporal drift on the 2024 holdout set, successfully mitigating severe class imbalances using balanced weighting.
  • •Automated hyperparameter optimization via Optuna and extracted deep model interpretability using SHAP (global, class-level, and local force plots) to identify key demographic drivers like age and household position.
  • •Stack: Python, CatBoost, SHAP, Optuna, Scikit-learn
    • okyeame-tts⁠, One Shot Voice Cloning Text-to-Speech Model
  • •Fine-tuned XTTS v2 on 30,000+ aligned Ghanaian English audio chunks from a 2,700-hour ASR dataset to build the first open-source TTS model trained exclusively on Ghanaian English speech.
  • •Built an end-to-end pipeline covering data collection, CTC forced alignment, Coqui format conversion, model training, and UTMOS-based evaluation.
    • 03/2026
  • •Deployed a live zero-shot voice cloning demo on HuggingFace Spaces.
  • •Stage 1 model achieves a best eval loss of 3.527 and produces distinctly Ghanaian-accented speech with cross-speaker voice cloning from short reference clips.
  • •Stack: Python · PyTorch · XTTS v2 · CTC Forced Aligner · HuggingFace · Gradio · Kaggle
  • •Architected and deployed a fully local, RESTful NLP microservice using FastAPI to classify user emotional states and generate empathetic, context-aware support messages.
  • •Fine-tuned a Qwen 1.5 (0.5B) Small Language Model (SLM) using LoRA (Low-Rank Adaptation) and 4-bit quantization to fit resource constraints, achieving a 0.70 overall accuracy and 0.69 weighted F1-score for 7-class macro-emotion classification.
    • 12/2025
  • •Integrated a Phi-3.5 Mini Instruct model for conditional natural language generation based on detected emotional intent.
  • •Engineered an end-to-end reproducible pipeline, including data preprocessing mapping 28 GoEmotions labels to 7 macro-emotions, custom tokenization with masking strategies, and rigorous model evaluation.
  • •Stack: Python, FastAPI, Qwen, Phi-3.5, LoRA, HuggingFace
  • •Engineered and deployed a cloud-native web application on GCP that provides end-to-end automated data cleaning for raw, unstructured user uploads.
  • •Designed a data processing pipeline utilizing Python to programmatically sanitize datasets, compute statistical summaries, and generate comprehensive Exploratory Data Analysis (EDA) reports in Excel.
    • 10/2025
  • •Integrated an automated email delivery system to asynchronously dispatch the cleaned data artifacts alongside three custom data visualization plots, drastically reducing manual preprocessing time for end-users.
  • •Stack: Python, GCP, R
    • 2 / 3
    District Report Automation
  • •Designed and implemented an automated data pipeline in R using R Markdown (Rmd) and Quarto (Qmd) to produce customized statistical reports for all 261 Metropolitan, Municipal, and District Assemblies (MMDAs) in Ghana.
  • •Engineered a scalable workflow that ingested administrative data at the district level, automating data extraction, cleaning, and visualization.
    • 09/2025
  • •Developed dynamic templates to generate tailored, high-quality reports with integrated visualizations (e.g., charts, tables) for each district, reducing manual effort by over 80% and improving report accuracy and delivery speed.
  • •Enabled data-driven policy decisions and development monitoring through efficient, reproducible, and transparent reporting processes.
  • •Stack: R, RMD, QMD, Latex
    • CERTIFICATIONS
    3 / 3