Overview
Pulse College's MSc Artificial Intelligence develops the mathematical, algorithmic and systems competencies required of modern AI practitioners. You will progress from probability and classical learning through deep architectures, language models, vision and sequential decision-making, finishing with MLOps evaluation and governance. Lab work uses contemporary Python toolchains and structured peer review. The capstone produces a documented AI system with performance, safety and monitoring artefacts suitable for industrial or research scrutiny — not a demo notebook alone.
MSc Artificial Intelligence is taught as one award, not a list of unrelated modules. The published length is 24 months, and the credit total is 36. Students move from Mathematics & Probability for AI through Mathematics & Probability for AI; Machine Learning Theory & Practice; Programming Intelligent Systems (Python); Data Engineering for ML Pipelines; Deep Learning & Representation Learning; Natural Language Processing & Large Language Models, and finish on MSc Artificial Intelligence Capstone Project.
Each course has a question, a method, and a submission. In this field, students define a measurable question, fit a method that can be checked, and report where it fails. Reading is a dataset description, a methods note, and one published evaluation. The artefact a marker expects is a project note with the question, the method, the result, and the limitation.
The award is built so that a graduate can do the following in practice: Design, train and evaluate supervised, unsupervised and deep learning models Build NLP and multimodal applications using modern transformer architectures Engineer reproducible ML pipelines with evaluation, monitoring and versioning
Teaching assumes the student can read a source, attempt a problem before the seminar, and revise after feedback. Attendance at live seminars is part of the design. The capstone or final course must use the methods of the earlier courses; a project that ignores them does not pass.
The pages for each course name the topics that are examined. Those topics are the syllabus. A brochure line is not a substitute for them.
The curriculum aligns with the technical depth expected at leading research universities and industry AI labs, with explicit attention to the EU GDPR, EU AI Act literacy and model documentation practices (model cards, data sheets). Optional GPU compute is available for distinction-level projects. Pathway exists into the PhD Data Science for outstanding graduates.
Who This Programme Is For
- ✦Software engineers transitioning into ML engineering or applied research
- ✦Data scientists seeking formal AI systems and deep learning depth
- ✦STEM graduates building competitive credentials for AI product roles
- ✦Technical founders embedding AI responsibly into products
Faculty spotlight
Dr. Yuki Tanaka leads representation learning research; Professor James Okello supervises responsible NLP for low-resource languages.
Programme Structure & Delivery
Coursework
Interactive modules, weekly readings, recorded and live seminars, and collaborative discussions with peers worldwide.
Assessment
Module assignments, proctored examinations, and capstone projects designed to demonstrate mastery of the field.
Learning Outcomes
- ✦Design, train and evaluate supervised, unsupervised and deep learning models
- ✦Build NLP and multimodal applications using modern transformer architectures
- ✦Engineer reproducible ML pipelines with evaluation, monitoring and versioning
- ✦Assess AI systems against fairness, safety and regulatory expectations
- ✦Communicate technical trade-offs to product and executive stakeholders
Career Pathways
- ✦Machine Learning Engineer
- ✦Applied AI Scientist
- ✦NLP / Computer Vision Engineer
- ✦MLOps Engineer
- ✦AI Product Technical Lead
Assessment
Modules are assessed through a published mix of coursework, applied projects, and examinations. Exam windows are announced in advance so students in other time zones are not forced into overnight sittings. Alternative arrangements are available where documented.
Curriculum
Term 1 — Foundations of Intelligent Systems
AI501Mathematics & Probability for AI- AI501-M1 · The quantity the later model will use
- AI501-M2 · A probability and the event it refers to
- AI501-M3 · An expectation and a case that breaks it
- AI501-M4 · A gradient or an update, and what it changes
- AI501-M5 · A numerical check a marker can repeat
- AI501-M6 · A result that is true only under a stated assumption
- AI501-M7 · Notation that must stay consistent
- AI501-M8 · The submission: a worked derivation and a numerical check
15 creditsAI502Machine Learning Theory & Practice- AI502-M1 · The prediction question and the unit of data
- AI502-M2 · A baseline the model must beat
- AI502-M3 · Features, leakage, and a split that is honest
- AI502-M4 · Error, a failure case, and who is harmed by it
- AI502-M5 · A simpler model and why it was not enough
- AI502-M6 · Monitoring after the model leaves the notebook
- AI502-M7 · What the model must not decide
- AI502-M8 · The submission: a method note, a result, and the limitation
15 creditsAI503Programming Intelligent Systems (Python)- AI503-M1 · The data structure the program uses
- AI503-M2 · A function and the test that would fail it
- AI503-M3 · A dataset read in, and a row that should be rejected
- AI503-M4 · A baseline implemented in code, not described
- AI503-M5 · A bug and the test that caught it
- AI503-M6 · What the program does not check
- AI503-M7 · Reproducibility of the run
- AI503-M8 · The submission: code, a test, and a short note
15 creditsAI504Data Engineering for ML Pipelines- AI504-M1 · The dataset, its source, and its unit
- AI504-M2 · A pipeline step and the failure of that step
- AI504-M3 · Leakage between training and later data
- AI504-M4 · A schema change and what breaks
- AI504-M5 · Quality checks before a model is fit
- AI504-M6 · A lineage note a marker can follow
- AI504-M7 · What the pipeline must not silently drop
- AI504-M8 · The submission: a pipeline note, a check, and a failure case
15 creditsTerm 2 — Advanced AI Methods
AI505Deep Learning & Representation Learning- AI505-M1 · A representation and what it discards
- AI505-M2 · A training objective and a baseline
- AI505-M3 · Overfitting, a split, and leakage
- AI505-M4 · An error analysis on cases the model misses
- AI505-M5 · Compute and data, and the cost of a higher score
- AI505-M6 · A failure mode a demo would hide
- AI505-M7 · What the model must not decide
- AI505-M8 · The submission: a model note, an evaluation, and the limit
15 creditsAI506Natural Language Processing & Large Language Models- AI506-M1 · The language task and the unit of text
- AI506-M2 · A representation of a sentence or document
- AI506-M3 · A metric and a case the metric flatters
- AI506-M4 · A baseline and the gain over it
- AI506-M5 · Bias in the text and in the output
- AI506-M6 · A use the system must not be given
- AI506-M7 · Error analysis on a sample a marker can read
- AI506-M8 · The submission: a task note, an evaluation, and the limit
15 creditsAI507Computer Vision & Multimodal AI- AI507-M1 · The visual task and the image or frame
- AI507-M2 · A representation and what it throws away
- AI507-M3 · Annotation and a label that is wrong
- AI507-M4 · A metric and a case it flatters
- AI507-M5 · Failure in ordinary lighting, occlusion, or scale
- AI507-M6 · A use the system must not be given
- AI507-M7 · Comparison with a simpler method
- AI507-M8 · The submission: an evaluation note and the limit
15 creditsAI508Reinforcement Learning & Sequential Decision Making- AI508-M1 · The sequential decision and the reward
- AI508-M2 · A policy and a baseline policy
- AI508-M3 · Credit assignment and a delayed reward
- AI508-M4 · Exploration and a cost of exploring
- AI508-M5 · A simulation and how it differs from the real system
- AI508-M6 · What the policy must not be allowed to do
- AI508-M7 · Instability and a run that does not repeat
- AI508-M8 · The submission: an experiment note, a result, and the limit
15 creditsTerm 3 — Systems, Governance & Capstone
AI601MLOps, Evaluation & Production AI Systems- AI601-M1 · The model that is leaving the notebook
- AI601-M2 · Data arriving in production and how it differs
- AI601-M3 · A monitor, a threshold, and a false alarm
- AI601-M4 · Rollback and the last known good model
- AI601-M5 · Who reviews a decision the model proposes
- AI601-M6 · A drift the metric did not catch
- AI601-M7 · The cost of a wrong automated action
- AI601-M8 · The submission: a production note with the monitor and the limit
15 creditsAI602Responsible AI, Safety & Governance- AI602-M1 · The decision the system is allowed to influence
- AI602-M2 · A harm, the person who bears it, and the evidence
- AI602-M3 · A metric that looks fair and is not
- AI602-M4 · A human review and when it can be skipped
- AI602-M5 · A record of the model and the data
- AI602-M6 · A use that is out of scope
- AI602-M7 · What this course does not treat as a law qualification
- AI602-M8 · The submission: a responsibility note on one system
15 creditsAI603MSc Artificial Intelligence Capstone Project- AI603-M1 · The prediction question and the unit of data
- AI603-M2 · A baseline the model must beat
- AI603-M3 · Features, leakage, and a split that is honest
- AI603-M4 · Error, a failure case, and who is harmed by it
- AI603-M5 · A simpler model and why it was not enough
- AI603-M6 · Monitoring after the model leaves the notebook
- AI603-M7 · What the model must not decide
- AI603-M8 · The submission: a method note, a result, and the limitation
30 creditsAdmission requirements
Academic entry
- Bachelor's degree from an accredited institution (minimum 2:2 or GPA 2.5/4.0)
- Strong quantitative background (mathematics, computer science, engineering, physics or related)
English language
- English proficiency: IELTS 5.5, TOEFL 46, or equivalent (waived for prior English-medium study)
Documents
- Statement of Purpose (500 words minimum) outlining your goals and fit for the programme
- Two academic or professional references with contact details
- CV or résumé documenting relevant education and professional experience
Programme questions
How long is the MSc Artificial Intelligence?+
The published duration is 24 months. Teaching language: English. Actual time-to-complete depends on mode and any recognised prior learning.
Can I study this from my country?+
This is a fully online award. You study from your country. No student visa and no campus relocation are required.
What does it cost?+
Degree tuition for this award is published as £0 / tuition-free on the online pathway. All online degree examinations and certificates are 100% free and sponsored — the only cost for online study is the application fee. Campus tuition and campus examinations are charged separately. Check the Fees page for details.
What are the entry requirements?+
Requirements are grouped on this page (academic, English, documents). Equivalent qualifications are considered. English may be waived after prior English-medium study.
How is the programme assessed?+
Assessment is typically a mix of coursework, projects, and examinations. Doctoral awards include a thesis or dissertation and an oral examination. Details sit in the programme specification and module outlines.
Will this award be recognised where I live?+
Recognition of the award for local employment, professional licence, or ministry attestation is decided by your employer or regulator. Pulse College publishes verification pages for certificates. We do not claim automatic equivalence in every country.
How do I apply?+
Start an application on this website. Progress is saved from the first step. Admissions: admissions@acapulse.com.