What does an efficient GTA hiring process look like?
The Graduate Teaching Assistant hiring process at Georgia Tech relied on manual spreadsheets, email chains, and paper forms. Administrative staff spent 8-12 hours per hiring cycle managing a fragmented workflow that resulted in frequent data entry errors, lost applications, and delayed onboarding.
I was brought in to redesign the entire process from end to end, with the goal of eliminating manual data entry while preserving the flexibility faculty and administrators need.
"How might we automate the hiring pipeline while maintaining the flexibility that faculty need to make nuanced hiring decisions?"
Managing Organizational Ambiguity in a Fragmented Workflow
With 20-30 GTA hires per semester across multiple departments, inefficiencies compounded into significant time loss and frustration. The challenge wasn't just technical, it was organizational: multiple stakeholders, unclear ownership, and resistance to process change.
Understanding the Problem Through Contextual Inquiry
I conducted contextual inquiry sessions with administrative staff, observing their actual workflow and identifying pain points in real-time. This hands-on approach revealed issues that interviews alone would have missed.
Research Methods
- Contextual Inquiry: 6 sessions observing staff in their environment
- Stakeholder Interviews: 8 participants across faculty and admin roles
- Process Shadowing: Full hiring cycle observation (3 weeks)
- Artifact Analysis: Review of forms, emails, and spreadsheets
Key Insights
Data Re-entry Was the Primary Error Source
The same information was entered manually 4-5 times across different systems. Each transfer introduced error potential. Staff spent more time checking data than entering it.
Status Visibility Was Non-existent
Faculty had no way to check application status without emailing administrators. This generated dozens of status-check emails per cycle, adding overhead for everyone.
Approval Bottlenecks Were Hidden
Applications sat in email inboxes waiting for approval. Without visibility, delays weren't surfaced until they became urgent problems.
An End-to-End Automated Pipeline with Built-in Validation
I designed and built an automated workflow using Microsoft Power Automate integrated with SharePoint, Forms, and DocuSign. The system eliminates manual data entry while preserving the flexibility faculty need.
Automatic Validation
Format checks and required field enforcement at point of entry eliminate downstream errors.
Real-time Status
All stakeholders can see application status without email threads or manual updates.
Automated DocuSign
Offer letters generated and sent automatically upon faculty approval.
Complete Audit Trail
Full history for compliance and accountability requirements.
From 23 Steps to Streamlined Flow
I mapped the existing workflow to identify redundancies and error-prone handoffs. The original process involved 23 discrete steps with 7 different tools and multiple manual data transfers.
The Redesigned Process
The new workflow consolidates all steps into a single automated pipeline. Data flows from application to onboarding without manual re-entry.
Visualizing the Hiring Journey
I created detailed storyboards to illustrate the pain points in the existing hiring process and how the automated system would transform the experience for administrative staff.
Measurable Impact on Administrative Efficiency
The system was deployed for the Spring 2026 hiring cycle. Results were measured against the previous three semesters.
What I Learned
-
Process Change Is Harder Than Technology Change
The technical implementation was straightforward. Getting stakeholders to adopt new workflows required careful change management, pilot programs, and visible wins to build trust.
-
Contextual Inquiry Reveals Hidden Work
Interviews told me what people thought they did. Observation showed me what they actually did, including workarounds and undocumented processes that were critical to understand.
-
Design for Edge Cases Early
The happy path was easy. The exceptions, cancellations, mid-process changes, and manual overrides, were where the real complexity lived. Building for these from the start prevented rework.