⭐ Featured Project

Instructor & Learner Insights Automation

Built a privacy-first analytics and AI pipeline that converts learner survey data into actionable insights, enabling instructors to plan classes based on evidence rather than intuition.

Technologies & Tools

PythonLLMSalesforceData Privacy

📊 Impact: Privacy-first AI insights for data-driven instruction

Problem

Instructors at Tech Goes Home rely on learner feedback and outcomes data to adapt their teaching, but historically this data was either too raw, too delayed, or too difficult to interpret. As a result, instructional adjustments were often based on intuition rather than evidence. Learner survey data lived in Salesforce but was underutilized, instructors lacked time and tools to analyze raw data, reports focused on metrics not interpretation or action, and instructional decisions were often based on experience rather than evidence.

My Role

I owned the solution end-to-end, including designing a privacy-safe analytics architecture, defining PII-stripping and data minimization rules, building the Salesforce → Python analytics pipeline, integrating LLMs to generate human-readable insights, and designing instructor-facing reports aligned to teaching workflows. This was not a reporting enhancement — it was a decision-support system.

Solution

Designed an automated insights pipeline that transforms learner survey data into structured, actionable insights for instructors. The system securely analyzes de-identified learner data using Python-based analytics and LLM-driven insight generation, then delivers concise, instructor-ready reports that directly inform lesson planning and instructional strategy.

Architecture

High-Level Data Flow

1

High-Level Data Flow: (1) Learner survey data extracted from Salesforce

2

Personally identifiable information (PII) is stripped

3

De-identified data securely sent to Python analytics service

4

Statistical patterns, trends, and deltas computed

5

Aggregated results passed to LLM for insight generation

6

Structured, instructor-ready report generated and delivered. Used Salesforce as source system, privacy layer with PII removal and field-level filtering, Python analytics engine, and LLM-based insight generation.

Key Design Decisions

🔹Privacy-by-Design - Removed PII before any external processing, ensuring instructors received aggregated insights not individual responses
🔹Insights Over Metrics - Focused on what instructors should do, not just what changed, structured to align with lesson planning
🔹AI as Interpreter - LLM used to translate analysis into insights, not invent conclusions, grounded in computed analytics

Results

  • Enabled instructors to adjust class plans based on real learner needs
  • Increased adoption of survey data in instructional planning
  • Reduced manual analysis and interpretation effort
  • Improved alignment between learner feedback and course delivery
  • Established scalable framework for AI-assisted educational insights

Technologies Used

PythonLLMSalesforce APIData PrivacyStatistical Analysis