AllDataSI

Lab · Agentic AI · Local LLMs & RAG · Enterprise Data

Ridgeline

An assistant that answers a signed-in student's questions about their own progress. It reads their records in PeopleSoft Campus Solutions and Canvas, cites the university's policy handbook, and runs on either Claude or a self-hosted model. The privacy rules live in the database, not in the prompt.

PostgreSQL 16pgvectorHybrid RAGClaude Opus 5.5Ollama · vLLMFastAPIDockerFERPA · GDPR
Ridgeline answering a student: a table of this term's classes with Canvas scores, a note on late work, three assignments due Nov 16, and a pointer to the student's advisor.
A student asks how they're doing this term. The answer combines Campus Solutions enrollment, Canvas scores and upcoming assignments, and names their advisor.

How it works

Two sources, one answer

Every answer draws on the student's own records and the university's written policy, and shows which of each it used.

The How I answered panel: model, time, a Cloud badge explaining where the data went, and three record lookups with their source system, field count and timing.

How I answered

Under each answer: the model, the time it took, where the data went, every record lookup with its source system, and the handbook sections it retrieved.

Handbook section AR-3 on academic standing and probation, opened from a citation on a phone.

Citations open the source

Answers cite handbook sections such as [AR-3]. Tapping a citation opens the policy text itself, so the student can check it.

Privacy by construction

The model can't ask for someone else's records

  • No IDs in tools. Every tool takes only the opaque session id, so neither the model nor the code can request another student's records.
  • Function-only access. The agent's database login can execute the student_portal functions and read no tables.
  • Consent and audit. Nothing runs before the student accepts the privacy notice. Every access is logged by category and time, never by content.
  • Minimised output. Tool output leaves out IDs, national IDs, birth dates, addresses, classmates and health details. Conversations are never stored.
  • Student controls. Export a copy of your data without sending it to a model, read the notice again, or withdraw consent and sign out.
The privacy notice shown before first use, with No thanks and I agree buttons.

The notice names the controller, the processor, the legal basis and the FERPA exception it relies on. A compliance map ties each FERPA, GLBA and GDPR requirement to the control that meets it, and SQL tests check the isolation, consent, minimisation and audit guarantees.

Cloud or campus

Same tools, switchable model

The Choose the AI model dialog with Claude Opus 5.5 on the Anthropic API and Qwen 2.5 7B self-hosted.

The student picks the model live. The tools, the privacy rules and the records stay the same; only the model that writes the answer changes.

Embeddings always run locally (bge-small-en-v1.5, baked into the image), so the handbook and questions are never sent anywhere to be embedded. For the self-hosted model, thinking is off and handbook sections are retrieved before generation, which smaller models handle more reliably than searching for themselves.

The self-hosted model doesn't need a data centre. On a mini PC's built-in Radeon 780M graphics, Gemma 4 12B starts answering in under ten seconds and finishes in about twenty. The answer streams in as it's written, so the student reads while the model writes. The Claude answer above took 11 seconds.

Gemma 4 12B (Q4) on a Ryzen 7 8745HS mini PC with Ollama, same student, one run each. "Text appears" is when the first words of the answer show on screen.
QuestionCPU only, before tuningCPU, streamed and trimmedRadeon 780M graphics
"How am I doing in my classes this term?"141 stext appears at 26 s, done at 46 stext appears at 9.5 s, done at 20 s
"What happens if my GPA drops below 2.0?"73 stext appears at 18 s, done at 33 stext appears at 5.6 s, done at 17 s

The tuning: the model reads the fixed system prompt and tool definitions once at startup and keeps them cached, fewer handbook sections go into each prompt, and record lookups are sent as compact JSON. The integrated graphics read prompts about five times faster than the CPU (about 150 tokens a second against 30). Writing speed barely changes (about 14 tokens a second against 11), because the graphics share the CPU's memory and writing is limited by memory speed. A dedicated GPU or a data-centre endpoint (vLLM, TGI, NIM) speeds up both.

The roots

A university in one PostgreSQL database

Ridgeline sits on a teaching database for a fictional Summit Ridge University, filled with deterministic synthetic data. It is shaped like the real systems but isn't Oracle PeopleSoft or an Instructure export.

LayerWhat it modelsScale
PeopleSoft HCMEmployees, effective-dated job history, positions, payroll and security77 employees, 12 years of history, 3.5k paychecks
Campus SolutionsAdmissions, programs, class schedule, enrollment, grades and GPA, degrees, billing, aid and holds200 students, 256 sections, 2.9k enrollments
Canvas Data 2The LMS warehouse as dap syncdb lays it out, fed from Campus Solutions on SIS ids272 courses, 3k assignments, 29k submissions
AnalyticsRoster sync exceptions, grade reconciliation and an at-risk engagement flagCross-system views
Student portalPrivacy-scoped functions, consent, audit log and the policy vector storeThe only surface the agent can touch

Run it

A local demo, by design

Ridgeline runs with Docker Compose on one machine: PostgreSQL with pgvector, the web app, and optionally Ollama. It isn't hosted here, because a real deployment needs a vendor contract, SSO and a DPIA first.

# Claude backend: put ANTHROPIC_API_KEY in .env
./scripts/demo.sh            # http://localhost:8000
./scripts/demo.sh --local    # also enable the self-hosted model

# every tool and the retrieval, with no model calls
docker compose --profile agent run --rm agent --self-test

Every name, grade and record in the demo is synthetic. There is no public sign-in; the screenshots on this page come from a local run.