Kelmer Fernandes

MSc Artificial Intelligence candidate at the University of Nottingham with a First‑class BSc in Computer Science (AI) and the NTU Comp Sci Department Prize recipient. Proficient in Python, PyTorch, TensorFlow, scikit‑learn and built full ML systems involving data pipelines, model optimisation, and scalable CI/CD MLOps deployments on AWS/Docker. Led a research project improving predictive accuracy by 12% and reducing inference latency by 30%, delivering an enterprise-ready, governance-compliant solution. Hands-on experience with big data processing (PySpark, Hadoop), NLP/LLM systems, including prompt engineering and vectorised search, fuzzy logic, and ethical AI development. Strong analytical and problem-solving skills, clear communication with stakeholders, and adept at integrating AI models into robust production pipelines. Strong foundation in statistics and SQL, quick to master new frameworks. Passionate about exploring real-time analytics, generative AI, and AI governance in practical environments.

• Automated AI Impact Assessment Tool (MSc Summer Research): Developed a React/TypeScript frontend and Python backend integrating LLMs to guide AI developers through ICO/AIIA-aligned impact assessments, delivering collaborative reporting features via user testing iterations.
• VirtualDoc Healthcare Chatbot ( MSc Human-AI Interaction): Built a full-stack chatbot using NHS APIs, scikit-learn (TF-IDF + Naive Bayes), NLTK, and fuzzy matching to classify symptoms and enable appointment booking with JSON persistence and intent recognition.
• Distributed Density-Aware Oversampling (MSc Big Data): Engineered PySpark pipelines for a 100M-row phishing dataset, implementing KD-tree-based oversampling and benchmarking DAO variants for 3.2× speedup with visualised scalability metrics.
• E-commerce Price Comparison Tool (BSc Final Year Project): Created a React/TypeScript + Flask web app scraping product data from Amazon/eBay/Argos, with AWS-backed MySQL persistence, ML-based deal recommendations, and user-tested dashboards.
• Forecasting Sea-Level Rise in London (MSc Data Science with ML): Modelled temperature projections using Random Forest and XGBoost (MAE = 0.26°C), converting outputs into SLR forecasts under four scenarios to inform long-term Thames Barrier policy planning.