Browse categories
Explore
Fiverr Pro
English
$
USD
Kirmani Oussema
@oussemakirmani
Tunisia
English, Arabic, French, German
About me
Computer Science Engineer from ENSI with expertise in Python, Pandas, NumPy, machine learning, and deep learning (TensorFlow, PyTorch, Keras). Currently pursuing a Master’s thesis in Sustainable AI at Algoma University. Conducted HAR research using Transformers and CNN-BiLSTM at Acadia University (MITACS), developed a biomedical NER system(F1>80%), and automated testing, reducing manual effort by 40%.Passionate about leveraging AI, I excel at data analysis, visualization, and predictive modeling, turning complex datasets into actionable insights and sustainable AI solutions.
Skills
Kirmani Oussema
Offline •
Average response time: 1 hour
See my services
Data Analytics Consultation
I will clean and preprocess your dataset in python
Work experience
Graduate Research Assistant
Algoma University • Part-time
Jan 2026 - Present • 4 mos
Conducted a systematic literature review (SLR) on Green AI and AI sustainability in Large Language Model (LLM) code generation Extracted and analyzed research findings aligned with 15 structured research questions related to energy-efficient AI and sustainable code generation. Identified and synthesized insights regarding prompting strategies and benchmarking methods used in evaluating LLM-generated code. Contributed to evidence synthesis and thematic analysis to support research outcomes in Green AI and sustainable AI systems
Graduate Teaching Assistant
Algoma University Faculty of Computer Science & Technology • Part-time
Jan 2026 - Present • 4 mos
Winter Term (January 2026 – April 2026) Created and graded assignments in JavaScript, HTML, and CSS. Delivered online lectures covering core front-end development concepts. Recorded instructional videos explaining JavaScript, HTML, and CSS, including guided walkthroughs and examples for students.
AI Research Engineer
Acadia University • Full-time
Jun 2025 - Sep 2025 • 3 mos
Deep Learning for Human Activity Recognition – Acadia University Developed deep learning models for human activity recognition using two distinct datasets. Sub-Project 1: UCHAR Dataset Designed and implemented a CNN-LSTM architecture for activity classification. Achieved 93.67% accuracy in predicting six activity classes. Conducted data preprocessing, cleaning, and feature engineering to enhance model performance. Sub-Project 2: NTU RGB+D 60 Skeleton Dataset Developed a CNN×LSTM architecture to recognize human actions from skeleton data. Reached 79.45% accuracy after optimizing the model and preprocessing steps. Performed model evaluation, architecture optimization, and accuracy assessment for both datasets. Applied end-to-end deep learning pipeline development from raw data to final performance metrics.
