LLM Training and Data Collection
AI/ML Data Collection Moderator & UX Research - Apple Inc. via INSPYR Solutions
In these user studies, we encourage conversational engagement through effective use of tone, active listening, and adaptive communication to help train an LLM model through various exercises that targeted different areas of the Model's stressors.
-
Lead over 100 natural, multi-tier conversations between participants and AI agents to simulate real-world dialogue, in roughly 3 month time frame.
-
Created supportive environments where participants feel at ease expressing emotion and speaking naturally, to successfully gather Authentic, Natural data that was used to flag hundreds of specific events to the developers.
-
Through open-ended and ambiguous scenarios and guiding conversations without influencing participant response, the team was able to identify errors to inspire rapid diagnosing through direct communication with the developers.
Thus resulting in solutions that help to improve the over experience for the user during and after testing.
Revolutionizing
Familiar Tools
A Virtual Assistant and Chat Bot for users to “Write with Ai”.
As a Data Collection Moderator for Apple, we tested various Beta versions of the LLM with over 100 participants to prime the model for market.
Conducting a User Study
.png)
Select the Info Graphic to Learn More
IOS &
Device Management
Select the arrows to view
AI/ML Calibration
Cross Collaboration with Developers
.png)
Data Evaluation and Heuristics
Every Project is an Opportunity to Grow
Key Take -Aways
-
Refined my ability to work quickly and accurately under pressure.
-
Improved my ability to provide clear, detailed documentation.
-
Strengthened my confindence in asking for help.