Louise Ferreira (Luli)

Artificial Intelligence Researcher | Machine Learning Specialist | Cybersecurity Professional
Transforming technical innovation into equitable AI systems at the intersection of NLP, fairness, and privacy

Technical Expertise

• Machine Learning & AI: Transformer architectures, fairness-aware NLP, recommender systems, BERT/RoBERTa fine-tuning
• Algorithm Development: Bias detection/mitigation, explainable AI (SHAP, LIME), clustering algorithms (K-means, SVD)
• Technical Stack: Python, PyTorch, HuggingFace, Scikit-learn, Pandas, Fairlearn, AIF360
• Cybersecurity: Data Loss Prevention (DLP), privacy engineering, open-source intelligence (OSINT)

Research Focus

Research Areas

Fairness-Aware NLP Developing transformer-based systems with built-in bias detection and mitigation.

Intersectional AI Ethics Creating models that account for overlapping dimensions of identity in data representation.

Privacy-Preserving ML Developing techniques that maintain data privacy while ensuring algorithmic fairness.

Featured Projects

Code Snapshot: Fairness-Aware ML

Academic Journey

2024-2025	➤	MSc in AI and its Applications University of Essex Chevening Scholar
⋮
2014-2016	➤	MA in Cultural Studies Jagiellonian University Transatlantic Studies
⋮
2009-2013	➤	BA in International Relations Federal University of Santa Catarina

GitHub & Development Metrics

Research & Publications

Intersectional Approaches to AI Safety A framework for evaluating safety systems across multiple dimensions of identity Expected: Winter 2025 Final Dissertation (2025) Topic: Detecting and Mitigating Algorithmic Bias in ML Models Categorizing Adult Content Metadata Focus: Black women and intersectional identities (e.g., BBW, Trans, Interracial) Dataset: 134,350 videos, categorized (asian, latina, interracial, ebony, bbw, brazilian, indian, arab, transgender) Metadata Fields: Tags, views, ratings, duration, category labels Techniques & Models: Descriptive statistics on metadata and tag frequency Bias and fairness analysis across groups Fairness metrics (e.g., Equal Opportunity, Demographic Parity) Text classification, unsupervised clustering, counterfactual data augmentation Text Analytics Project: Sentiment Analysis on Amazon Reviews (NLP) Models Used: VADER – rule-based sentiment analysis DistilBERT – transformer-based classification RoBERTa – advanced transformer for nuanced detection Logistic Regression – baseline supervised model Tasks: Text preprocessing (tokenization, lemmatization, stopword removal) Evaluation on unbalanced data (precision, recall, F1, confusion matrix) Comparison of unsupervised, supervised, and transformer-based models Information Retrieval Project: Analyzing Denzel Washington's Career Tools Used: Elasticsearch, Kibana Methods: Indexing & querying IMDb data Search relevance tuning (boosting by tags or roles) Visual dashboards of genre, co-stars & critical reception Programming Language: Python Topics: Basic syntax, loops, functions, data structures, OOP AI & Machine Learning with Applications Models & Topics: K-Means Clustering SVD (latent factor modelling) ANNs for classification & regression Hybrid recommender systems Evolutionary feature selection Imbalanced data metrics (AUC, Precision@K, NDCG)

Technical Proficiency

Connect

Bringing together technical excellence, lived experience, and interdisciplinary perspectives
to create more equitable and representative AI systems

Expertise & Certifications

---

🖼 Icon Credits

• Brain icons created by juicy_fish - Flaticon
• Diversity icons created by Freepik - Flaticon
• Privacy icons created by Flat Icons - Flaticon