Database Fundamentals
Databases have become the grounding substrate of AI systems: retrieval quality, more than model choice, determines whether a generative application answers from authoritative records, and agents now both draft SQL and query production data directly. The course therefore keeps its relational core intact while adding vector and hybrid retrieval, disciplined verification of machine-generated SQL and migrations, and least-privilege governance of agent access to data.
Current catalog prerequisites — (CSC 200 or 2000).
Current description → proposed description
This course provides students with the background to plan, design, implement, maintain, and use database management systems. It addresses database structures, requirements, functions and evaluation of database management systems. The course focuses on the relational database model, standard SQL language, database structure normalization, conceptual data modeling, and the entity-relationship data model. Concepts of data integrity, security, privacy, and concurrence control are included.
Prerequisites: (CSC 200 or 2000).
This course provides students with the background to plan, design, implement, secure, and govern database systems that serve both human users and autonomous software agents. It addresses conceptual data modeling, the entity-relationship model, the relational model, normalization, standard SQL, transactions, and concurrency control, and extends those foundations to the storage and retrieval patterns on which modern AI systems depend. Students work with vector and hybrid search alongside relational queries, treat embeddings, chunk boundaries, and metadata as deliberate design decisions, and build retrieval layers that ground generative systems in authoritative records. Because queries, schemas, and migrations are now routinely drafted by AI coding assistants, the course emphasizes verification: reading execution plans, exercising constraints with targeted test data, and confirming that generated SQL is correct, performant, and safe before it reaches production. Data integrity, privacy, least-privilege access for agent tooling, injection exposure, retention, and auditable data lineage are treated as core professional obligations and as Christian stewardship of information entrusted to the developer.
What changes
- Verification of AI-drafted SQL, schemas, and migrations
- Vector and hybrid search added beside the relational model
- Database as the grounding layer for retrieval-augmented systems
- Least-privilege credentials and injection exposure for agent tooling
- Provenance, retention, and auditable lineage as design requirements
6 proposed outcomes, mapped to 9 program outcomes
Each outcome below is written to be observable and assessable, and each is mapped to the program learning outcomes for which it produces evidence.
Students will be able to design a normalized relational schema and entity-relationship model by deriving functional dependencies from sample data instances and logged application and agent query patterns alongside ambiguous stakeholder requirements, documenting the trade-offs behind each key, constraint, and normal-form decision.
Inferring functional dependencies from sample data instances and from logged application and agent query patterns, then reconciling them with ambiguous stakeholder statements to justify a normal form, is the combined quantitative and qualitative analysis PLO 6.2 names and the meaningful conclusion it asks students to draw.
Students will be able to evaluate SQL, schema changes, and migrations drafted by AI coding assistants for correctness, performance, and safety, using execution plans, targeted test data, and constraint checks as evidence, and revising a personal verification checklist based on the defects each review surfaces.
Judging a generated query's result set, plan cost, and constraint violations against measured evidence is the critical analysis of data-driven outcomes in PLO 4.1, while revising a personal verification checklist from the defects each review surfaces is the continuous self-assessment of one's own problem-solving approach that PLO 4.3 describes.
Students will be able to implement and evaluate a retrieval layer that combines relational queries with full-text and vector search to ground a generative AI application in authoritative institutional records, comparing ranking strategies and explaining why a given record was retrieved.
Comparing competing ranking strategies over a shared query set and recommending one is the proposal and evaluation of AI-driven solutions in PLO 4.2, and explaining why a given record surfaced from its similarity scores, filters, and fused ranking is the interpretation and explanation of AI data outcomes in PLO 6.1.
Students will be able to analyze transaction isolation and concurrency-control behavior in workloads where autonomous agents and human users write to the same records, identifying the anomalies each isolation level permits and assessing the downstream harm when an autonomous writer corrupts shared records.
Naming the lost updates, write skew, and phantom reads an isolation level permits under measured concurrent load and then assessing who is harmed downstream when a high-frequency autonomous writer corrupts shared state is the critical examination of data-driven outcomes and their ethical and societal impact that PLO 4.1 names.
Students will be able to critique the access-control, privacy, retention, and audit posture of a database exposed to AI agent tooling, including least-privilege credentials, injection pathways, and the handling of personally identifiable data.
Examining privacy, transparency, and accountability in how a database is exposed to agent tooling is exactly the professional ethical reflection named in PLO 1.2, and judging who is harmed when personally identifiable data leaks through an agent tool is the fairness and moral-impact assessment of PLO 6.3.
Students will be able to articulate the structure, limitations, and provenance of a data set to non-specialist stakeholders in a written data dictionary and an oral briefing.
Producing a written data dictionary and delivering an oral briefing is the transparent written and spoken communication of PLO 5.1, and stating a data set's limitations and provenance in terms a non-specialist can act on is the translation for broader audiences described in PLO 5.3.
Program outcomes this course reaches
Filled cells are program learning outcomes with at least one supporting course learning outcome in this course. Sparse coverage is expected — no single course carries all eighteen.