Intelligent Learning Case study
Sagax
Educational data-science products for transcript and syllabus analysis, skills mapping and academic decision support.
Intelligent Learning
Sagax Sagax
Lead Engineer
System map · Processing pipeline
Engineering approach
- Scored, not decided Evidence for an advisor, who carries the academic consequence
- Skills, not titles Differently worded courses held against each other
Analysis path
Transcript or syllabus
Free-form, no shared schema
Document analysis
A structured representation
Skills mapping
What was taught, not the title
Equivalency scoring
With supporting evidence
Decision support
The advisor decides
Engineering
Document analysis · Skills mapping · Equivalency scoring · Institutional integrations
Public product · Proprietary source code
- Role
- Lead Engineer
- Disclosure
- Public product · Proprietary source code
- Links
- Product website
The problem
What the system had to solve.
A transcript says what a student has done; a syllabus says what a course covers. Deciding whether one satisfies the other drives transfer credit, admissions and advising — and it is answered by a person reading both and forming a judgement.
The documents are unstructured and institution-specific. Two courses with the same title can cover different material, and two with different titles can be equivalent.
What I owned
The responsibility behind the work.
Led product development across the core application architecture, user-facing functionality, administrative workflows and the intelligent educational-data features underneath them.
Engineering approach
How it was built.
Transcripts and syllabi are analysed into structured representations before anything is compared, and mapped to skills so two differently worded courses can be held against each other on what they taught rather than on their titles. Equivalency is scored with supporting evidence rather than decided, because an academic ruling has consequences for a student and has to be explainable and appealable.
Analysis path
- Document
- Structure
- Skills
- Score
- Advise
Two documents that never agree
What academic decision support actually has to solve.
A transcript says what a student has done. A syllabus says what a course covers. Deciding whether one satisfies the other is the question behind transfer credit, admissions, advising and programme planning — and it is currently answered by a person reading both and forming a judgement.
The documents are unstructured, institution-specific and inconsistent. Two courses with the same title can cover different material, and two courses with different titles can be equivalent. That is a data problem before it is a product.
Why it resists automation
- Free-form documents with no shared schema
- Institution-specific vocabulary
- Titles that do not indicate content
- Judgements that carry academic consequences
From document to comparable structure
The step everything else depends on.
Nothing can be compared until both sides have been turned into structured, comparable representations. Analysis of transcripts and syllabi produces that structure, and skills are the common vocabulary that lets two differently worded courses be held against each other at all.
Analysis path
- Transcript or syllabus
- Document analysis
- Structured representation
- Skills mapping
- Equivalency scoring
- Decision support
Scoring rather than deciding
Where the system stops.
Course equivalency is scored, not decided. An academic decision has consequences for a student’s time and money, and the appropriate output is evidence and a score that an advisor acts on — not an automated ruling that someone has to work out how to appeal.
This is the same principle as any consequential automation: the system does the work that does not scale, and the human keeps the decision that carries the consequence.
Serving institutions, not just students
Administrative workflows and integrations.
The product has to fit institutional processes rather than replace them, which means administrative workflows for the staff who own these decisions and integrations with the systems that already hold the records. A tool that requires re-keying transcripts is a tool that does not get used.
Product surface
- Core application architecture
- Transcript and syllabus analysis
- Skills mapping and equivalency scoring
- Academic decision-support workflows
- Administrative and institutional integrations
Engineering scope
What I worked on.
- Core application architecture
- Transcript and syllabus analysis
- Skills mapping and course-equivalency scoring
- Academic decision-support workflows
- Administrative and institutional integrations
Why these decisions mattered
The trade-offs behind the product.
Skills as the common vocabulary
Score, do not decide
Structure first
Integrate with institutional systems
Impact
What the product changes.
The work that does not scale — reading and structuring documents — is automated, and the decision that carries academic consequence stays with an advisor, supported by evidence.
Technology
The technical surface.
- Educational data science
- Document analysis
- Skills mapping
- Decision support