Advanced Database and Web Development
In industry, database-backed web applications are now largely scaffolded by AI coding assistants and are increasingly consumed by software agents as well as by people, shifting the developer's work toward specification, retrieval design, review of generated code, and a vulnerability surface that now includes prompt injection and over-permissioned tool endpoints. The course keeps its web, data, and media identity while adding grounded retrieval, verification of work the developer did not type, and agent-era security and provenance practice.
Current catalog prerequisites — (CSC 430 or 3030).
Current description → proposed description
Databases are the primary information repositories of 21st-century information technology. This course focuses on web technologies, information architecture, digital media, web design and development, vulnerabilities, and social software.
Prerequisites: (CSC 430 or 3030).
Databases remain the primary information repositories of twenty-first-century information technology, and this course extends the relational foundations of CSC 3030 into the design, construction, and deployment of data-driven web applications. Students study web technologies, information architecture, digital media, accessible interface development, and social software, building applications whose data layer serves interactive users, published APIs, and the software agents that increasingly consume them. Retrieval is treated as an architectural concern, combining relational queries with vector and hybrid search so that any generative feature remains grounded in authoritative records. Because application code, schema migrations, and interface components are now routinely drafted by AI coding assistants, the course emphasizes specification, review, and verification of work the developer did not personally type. Vulnerabilities are studied in their current form, including injection, broken access control, cross-site scripting, prompt injection through untrusted content, over-permissioned agent endpoints, and dependency supply-chain risk. Privacy, provenance, content moderation, and data retention are treated as professional obligations and as Christian stewardship of information entrusted to the developer.
What changes
- Retrieval as architecture: relational queries joined with vector and hybrid search
- Data layer and APIs designed for agent clients as well as human users
- Review and verification of AI-drafted code, migrations, and components
- Agent-era vulnerabilities: prompt injection, over-permissioned endpoints, supply chain
- Provenance, moderation, and retention framed as stewardship
6 proposed outcomes, mapped to 8 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 database-backed web application whose data layer serves interactive users, published APIs, and agent clients, integrating relational queries with vector or hybrid retrieval so that generated content stays grounded in authoritative records.
Architecting a data layer that joins relational queries with vector or hybrid retrieval, and then demonstrating that generated content traces back to authoritative records, is proposing and evaluating an AI-driven solution to a complex problem, which is what 4.2 names.
Students will be able to evaluate AI-generated application code, schema migrations, and interface components against correctness, performance, accessibility, and security criteria, deciding on that evidence whether to accept, revise, or reject each contribution before deployment.
Judging machine-produced code, migrations, and components against measured correctness, performance, accessibility, and security evidence rather than accepting them on appearance is exactly the critical analysis of data-driven outcomes that 4.1 describes.
Students will be able to implement layered defenses for a database-backed web application against injection, broken access control, cross-site scripting, server-side request forgery, prompt injection through untrusted page content, over-permissioned agent tool endpoints, and dependency supply-chain compromise, analyzing for each defense which untrusted-input path it closes and who is harmed when it fails.
The outcome itself requires an analysis, for each defense, of which untrusted-input path it closes and who is harmed when it fails, so the assessable work includes reasoning about AI-specific failure paths such as untrusted page content redirecting the application's generative features and over-permissioned agent tool endpoints acting on that redirection, together with the resulting societal harm, which is the critical analysis of data-driven outcomes and their ethical implications named in 4.1.
Students will be able to construct information architecture and digital media that are legible to both human readers and machine consumers, using semantic markup, structured data, accessible navigation, content APIs, and explicit provenance and attribution metadata for AI-generated or AI-edited assets.
Labeling which assets were machine-generated and carrying that attribution through the interface is presenting information transparently and with integrity (5.1), and semantic markup, accessible navigation, and documented content APIs are what make technical material usable by non-specialist audiences and the tools they rely on (5.3).
Students will be able to interpret telemetry from a deployed application, including query execution plans, retrieval groundedness and citation accuracy, user task completion, and the latency and cost budgets of model calls, to identify which revisions the evidence supports.
Explaining what retrieval groundedness and citation-accuracy results actually say about a deployed AI feature is data interpretation in an AI setting (6.1), and identifying which revisions are supported by execution plans, task-completion rates, and latency and cost measurements is the quantitative analysis 6.2 requires.
Students will be able to critique the social consequences of the systems they build, including engagement-optimizing recommendation, moderation and labeling of AI-generated content, user consent, and data retention, framing their own design choices as stewardship of the data and attention entrusted to them.
Framing retention, consent, and disclosure decisions as stewardship of what users have entrusted to the developer is Christian conviction shaping professional judgment about privacy, transparency, and accountability (1.2), and weighing who is advantaged or harmed by a recommendation or moderation policy is assessment of an AI system's fairness and moral impact (6.3).
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.