Profile
PhD researcher building fair, calibrated and explainable clinical risk prediction models on linked UK population health data, transitioning into machine learning engineering from a decade leading Computer Science education. Undertaking a part-time PhD at the University of Portsmouth on early detection of Type-2 Diabetes risk in South Asian populations, with data partnerships spanning Connected Bradford and UK Biobank. Holds an MSc in Computer Science and Data Science. Builds ML systems end to end — multi-model comparison, calibration analysis, subgroup fairness auditing with Fairlearn, SHAP explainability and deployment through FastAPI, Docker and MLflow — alongside retrieval-augmented systems with formal hallucination evaluation. A decade of teaching and leading Computer Science across UK and UAE schools brings unusual strength in making technical work legible to non-specialist audiences. Seeking machine learning engineering roles in healthcare AI, applied ML or EdTech across the UAE and remote UK.
Technical Skills
Programming: Python (advanced), SQL, R, JavaScript, Java, C++
Machine Learning: PyTorch, TensorFlow, scikit-learn, XGBoost, LightGBM, Hugging Face Transformers
GenAI & LLM: LangChain, LlamaIndex, RAG architectures, vector databases (pgvector, Pinecone), QLoRA fine-tuning, evaluation harnesses
ML Engineering: FastAPI, Docker, MLflow, Git, GitHub Actions (CI/CD), model serving, experiment tracking
Data Engineering: pandas, NumPy, Spark, Hadoop, ETL and data pipeline design
Visualisation: Matplotlib, Seaborn, Power BI, Tableau
Cloud: AWS (S3, EC2, Lambda); Microsoft Azure
Specialist: Clinical risk prediction, fairness auditing (Fairlearn, AIF360), model calibration, explainable AI (SHAP)
Doctoral Research
PhD Candidate, Computer Science (part-time) — University of Portsmouth, UK
2024 – present
Research: Early Detection of Type-2 Diabetes Risk using AI/ML in the South Asian Population.
- Developing calibrated, fair and explainable machine learning models for early Type-2 Diabetes risk stratification from routinely collected demographic, lifestyle and clinical data.
- Data partnerships with Connected Bradford and UK Biobank; comparative study design across Bradford localities with high and low South Asian population density to assess equity impacts.
- Methods span logistic regression, gradient-boosted trees and deep learning, with calibration analysis (Brier score, reliability diagrams) and subgroup fairness audits (equalised odds, calibration-by-group).
- Conducted part-time alongside a full-time teaching post, requiring research that is resource-aware and deployable in real clinical settings.
Technical Projects
T2D Risk Predictor — Fair, Calibrated and Explainable ML for Type-2 Diabetes
Python, XGBoost, PyTorch, Fairlearn, SHAP, FastAPI, Docker, MLflow | github.com/kesserk/t2d-risk-predictor | In progress
- Public, productionised risk prediction pipeline on open health data, aligned to the doctoral research and built to be inspected.
- Four-model comparison — logistic regression, XGBoost, MLP and tabular transformer — evaluated on AUROC, AUPRC, Brier score and calibration-by-subgroup.
- Subgroup fairness audits via Fairlearn and SHAP-based explainability; served through FastAPI and Docker with MLflow tracking and drift monitoring.
Diabetes Guidance RAG — Retrieval-Augmented Q&A over NICE NG28 and ADA Standards of Care
LangChain, pgvector, PubMedBERT, RAGAS, Langfuse, FastAPI | github.com/kesserk/diabetes-rag | In progress
- Hybrid retrieval combining BM25 with dense PubMedBERT embeddings over canonical diabetes clinical guidelines, with cross-encoder reranking.
- Evaluation harness using RAGAS plus a hand-curated clinical QA set, explicitly measuring faithfulness and hallucination rate rather than assuming correctness.
- Deployed with FastAPI and pgvector, instrumented with Langfuse tracing.
Learning Analytics Early-Warning Model
Python, scikit-learn, SHAP | Prototype
- Model unifying NGRT, CAT4, ALIS, TIMSS and internal school assessment data to flag learners who may need support, returning explainable reasons alongside suggested interventions.
MarkMate — AI-Assisted Marking and Feedback
LLM application | In development
- Tool generating structured, mark-scheme-aligned feedback to reduce marking time while holding consistency, designed around real assessment workflows and currently seeking teacher testers.
Professional Experience
Computer Science Teacher — Nord Anglia International School Abu Dhabi, UAE
August 2024 – present
- Played a key role in establishing and growing the Secondary Computer Science department, including resources, schemes of work and digital learning platforms aligned to Cambridge IGCSE and A Level frameworks.
- Led development of a future-focused Computer Science curriculum embedding problem-solving, creativity and digital fluency across key stages.
- Facilitated enrichment through the Nord Anglia–MIT collaboration, guiding students through hands-on STEAM challenges and global innovation projects.
- Founded and lead the school's Minecraft Education Club, using game-based learning to teach computational thinking, teamwork and design thinking.
- Work with the leadership team to promote digital citizenship and safe, purposeful use of technology across the school community.
Gifted & Talented Coordinator and Computer Science Teacher — Safa Community School, Dubai, UAE
January 2021 – August 2024
- Built Power BI dashboards tracking student performance across approximately 200 students and 6 subjects, replacing reporting workflows previously done manually.
- Automated data ingestion and cleaning pipelines connecting the school management system to visualisation platforms, reducing manual reporting time.
- Designed and delivered a hands-on Python and data science curriculum using Jupyter, covering supervised and unsupervised learning, model evaluation and basic deployment patterns.
- Mentored students on independent machine learning projects including image classification and regression tasks.
- Taught GCSE and A Level Computer Science to students aged 11–18 and contributed to curriculum design across Key Stages 3, 4 and 5.
Mathematics and Computer Science Teacher — One in a Million Free School, Bradford, UK
September 2018 – July 2020
- Deputised for the Head of Department, running the Mathematics department in his absence.
- Led the design and planning of a new computing curriculum, including introductory Python, algorithms and data analytics.
- Conducted data analysis of examination performance to identify focus areas, then created and implemented targeted intervention programmes.
- Delivered the Edexcel GCSE Mathematics syllabus across KS3 and KS4; conducted teacher observation, performance management and NQT mentoring.
Computer Science Teacher — Lightcliffe Academy, Halifax, UK
March 2018 – July 2018
- Delivered the Computer Science National Curriculum across Years 7–13, introducing the new KS3 curriculum, the OCR GCSE course and a BTEC Level 3 course in ICT.
Curriculum Leader, Technology — Trinity Academy Sowerby Bridge, UK
September 2017 – March 2018
- Managed the Technology department covering Computer Science, iMedia and Business Studies, supervising delivery across specialist and non-specialist teachers.
- Implemented intervention strategies and monitored attainment through assessment and target setting.
Computer Science Teacher — GEMS Wellington International School, Dubai, UAE
September 2015 – July 2017
- Taught Cambridge iGCSE Computer Science, iGCSE ICT and IB curriculum to students aged 11–18, and contributed to curriculum design across Key Stages 3 and 4.
Earlier: ICT Trainee Teacher (PGCE placements) — King James Secondary School, Knaresborough and Bury Church of England High School, UK | 2014 – 2015
Education
PhD Computer Science (part-time)
University of Portsmouth, UK | 2024 – present
Early Detection of Type-2 Diabetes Risk using AI/ML in the South Asian Population
MSc Computer Science and Data Science
University of Sunderland, UK | 2023 – 2024
Machine learning, time series analysis, big data technologies, deep learning with TensorFlow and PyTorch; dissertation applying Bayesian methods to predictive modelling on health-relevant datasets
PGCE Computing & ICT
Manchester Metropolitan University, UK | 2014 – 2015
Subject Knowledge Enhancement Course, Computing
Manchester Metropolitan University, UK | 2014
LLB Law
Leeds Metropolitan University, UK | 2006 – 2010
Professional Development & Memberships
- Member, British Computer Society (BCS).
- DataCamp Machine Learning Scientist with Python track completed; Machine Learning Engineer specialisation in progress.
- Preparing for Microsoft AI-300: Machine Learning Operations (MLOps) Engineer Associate (successor to the retired DP-100).
Additional
Languages: English (native), Urdu (basic spoken)
Interests: Football, running and cycling; has run several marathons including London and New York, fundraising for charity
Volunteering: Local tuition centres, supporting students in Years 5–11 with Computer Science and Mathematics
Last updated August 2026.