← All Openings

AI Research Intern – ML Calibration & Authenticity

R&D Department · Remote · Internship · 3 – 6 Months

About the Internship

We are looking for a motivated Machine Learning Research Intern to contribute to research on trustworthy artificial intelligence, model calibration, and AI text drift analysis.

The project explores how machine-learning systems and generative language models can become more reliable, transparent, and robust when evaluating predictions, confidence scores, and evolving content. Our focus is not only on model accuracy, but also on confidence calibration, uncertainty awareness, drift metrics, stress-testing, and content provenance.

This is a research-focused opportunity for students or early-career candidates interested in machine learning, deep learning, natural language processing, responsible AI, uncertainty quantification, or data science.

Requirements

Key Responsibilities

• Support development and evaluation of machine-learning models for calibration, text drift analysis, and content authenticity.
• Clean, preprocess, and analyze datasets involving text, model outputs, sequence data, or structured metadata.
• Build reproducible Python notebooks and scripts for data processing, experiments, and model evaluation.
• Train and evaluate baseline machine-learning or deep-learning models using PyTorch or similar frameworks.
• Assist with model reliability analysis, including calibration error, uncertainty estimation, token entropy, and distribution shift checks.
• Benchmark detection and calibration pipelines against adversarial inputs, distribution shifts, and out-of-distribution data.
• Create clear charts, evaluation tables, and technical documentation for research findings.
• Support literature reviews, experiment tracking, reproducibility checks, and manuscript preparation where appropriate.
• Participate in remote research meetings and provide concise progress updates.

Required Qualifications

• Currently studying, or recently graduated from, Computer Science, Data Science, Artificial Intelligence, Mathematics, Statistics, or a related field.
• Good Python programming skills.
• Familiarity with machine-learning fundamentals, including training/testing splits, overfitting, classification, regression, and evaluation metrics.
• Experience with Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn, PyTorch, TensorFlow, or similar tools.
• Basic understanding of statistics, probability, and model evaluation metrics such as accuracy, precision, recall, F1 score, or calibration curves.
• Ability to write organised, reproducible notebooks or scripts.
• Comfortable working independently in a remote and asynchronous environment.
• Strong written communication skills, attention to detail, and willingness to learn.

Preferred Skills (Not Required)

• Experience with PyTorch, Hugging Face Transformers, NLTK, SpaCy, or NLP processing frameworks.
• Theoretical understanding of model calibration techniques (e.g., temperature scaling, KANs, spline functions) or uncertainty quantification.
• Familiarity with generative AI evaluation metrics, token entropy, perplexity, or synthetic media analysis.
• Interest in responsible AI, model safety, statistical rigor, confidence estimation, or content provenance.
• Experience with Git, GitHub, Docker, experiment tracking, cloud notebooks, or research-code documentation.
• Prior academic research, technical writing, literature review, or publication experience.

What You Will Learn

• How to develop and evaluate machine-learning models for confidence reliability and drift monitoring.
• Practical approaches to model calibration, uncertainty estimation, token entropy analysis, and trustworthy AI evaluation.
• Data-preprocessing workflows for language model outputs, unstructured text, and sequential data.
• How to design reproducible stress-testing experiments against adversarial inputs and distribution shifts.
• How to interpret statistical results carefully and communicate research limitations clearly.
• Research practices including technical documentation, experiment tracking, literature review, and collaborative code development.

What You Will Gain

• Hands-on experience in applied machine learning, trustworthy AI, and model reliability research.
• Mentorship in research methodology, model evaluation, and reproducible technical workflows.
• Portfolio-ready notebooks, visualisations, code, and documentation contributions where appropriate.
• Exposure to model calibration, AI drift detection, and content provenance research.
• Potential acknowledgement or authorship opportunities where the level of contribution meets applicable research standards.
• A structured opportunity to develop skills relevant to research, AI engineering, data science, and graduate study.

Selection Process

1. Review of CV/resume, technical skills, and relevant project links.
2. A short online interview to discuss motivation, availability, machine-learning fundamentals, and learning approach.
3. A short Google Colab coding assessment involving Python data analysis and basic machine-learning evaluation.
4. Final discussion regarding start date, expected commitment, and internship goals.

The assessment is designed to evaluate Python fundamentals, data-cleaning ability, model-evaluation reasoning, reproducible workflow practices, and careful interpretation of results. Advanced calibration knowledge is not required.

How to Apply

Please submit the following:

• Your CV or resume.
• A short note explaining your interest in the internship.
• Links to GitHub, portfolio, Google Colab notebooks, Kaggle, or relevant projects, if available.
• Your expected weekly availability and preferred start date.

Note: This is a research and technical learning opportunity. Specific research methods, unpublished findings, exact datasets, internal evaluation procedures, and other confidential project information will be shared only with the selected candidate where necessary.

View full details and apply →