← Back to Portfolio

Portfolio › AI-Enabled Analytics

AI-Enabled Survey Sentiment
& Theme Analysis

Turn thousands of open-text survey responses into actionable insight – automatically. AI scores every comment for sentiment and discovers the themes people actually talk about, so leadership can see what’s driving satisfaction (or frustration) at a glance. No manual coding. No six-week research cycle. Just filterable, decision-ready dashboards powered by Snowflake Cortex AI and Power BI. Built on synthetic data as a proof of concept.

7

Days to Prep
the Data

1

Day to
Actionable Results

100%

AI-Driven
Insight Coverage

Snowflake Cortex AI Microsoft Fabric Power BI Synthetic Data Only

Data source

Synthetic dataset

750 procedurally generated student survey records across 9 schools, 4 questions, 2 years (2023–2024). No student records, no institutional data, no identifiable information. Modelled on QILT survey structure.

IP position

Zero institutional data used

Built on PMPC infrastructure using only synthetic responses. Demonstrates Cortex AI SENTIMENT() and COMPLETE() pipeline capability without requiring access to any institution's survey platform or student data.

Coverage

750 records · 4 questions · 9 schools

Q1 Teaching Quality, Q2 Student Support, Q3 Learning Resources, Q4 Campus & Facilities. Domestic/International split. Undergraduate/Postgraduate. SES, GOS, and GOS-L survey types.

Loading dashboard...

How the pipeline works

End-to-end architecture from survey data to executive dashboard - built inside existing government cloud infrastructure.

Microsoft Azure Government-mandated infrastructure · Australian data sovereignty · Azure Australia East
📋
01

Data Preparation

Microsoft Excel

Survey responses SES · GOS · GOS-L
⚙️
ETL

Extract Transform Load

Microsoft Azure

Ingest · Validate · Enrich Medallion architecture
🥇
04

Gold

Analytics-ready

🥈
03

Silver

Cleansed · Validated

🥉
02

Bronze

Raw ingest

🧠
05

Cortex AI

Snowflake

SENTIMENT() COMPLETE() −1.0 to +1.0 Theme discovery
3,000 responses analysed in <3 min
Data never leaves Australia · Cross-region inference enabled for Cortex AI
Microsoft Fabric Power BI Semantic model · Interactive dashboards · Row-level security
06

Semantic Model

Fabric Lakehouse DirectQuery to Snowflake No data copy
07

Access Tiers

Tier 1 · full dataset Tier 2 · school view Tier 3 · course view Login-controlled · SSO
08

Executive Dashboard

Filter by school, cohort Sentiment gradient Drill to verbatim
09

Outputs

Thematic insight Sentiment score Quantitative + qualitative

Executive benefits

Speed

Minutes vs weeks. Hours to build, minutes to refresh.

Consistency

Eliminates manual variability. Same model, every response.

Actionability

Quantified theme insights. Colour-coded by sentiment.

Need qualitative survey data analysed at scale?

Snowflake Cortex AI sentiment and theme extraction, Microsoft Fabric semantic model, Power BI dashboards with faculty-level filtering. Scoped, quoted, delivered.