Lessons from LLM-Based Grading and Voice-Based Simulation Systems in EdTech
A practical look at what really happens when LLMs move from demo to production: lessons from deploying automated grading and interactive simulation systems in real EdTech products.
As part of the AI Lightning Round at the Kiss the Future Meetup — powered by SLO AI — I shared hands-on lessons from building and shipping production-grade LLM systems in education. Fast. Focused. Practical.
Event Details
Date: Tuesday, February 17, 2026
Location: Evropska kuća kulture i nacionalnih manjina, Sarajevo
Event: Kiss the Future Meetup — AI Lightning Round (powered by SLO AI)
Talk Abstract
Title: Lessons from LLM-Based Grading and Voice-Based Simulation Systems in EdTech
What really happens when LLMs move from demo to production?
At Trivial Group, we design and deliver large-scale LLM grading, evaluation, and simulation platforms that power automated feedback systems in education and interactive, dialogue-based simulators used for training and assessment. In this talk, I shared hands-on lessons from deploying these systems in real EdTech products, tackling challenges like:
- Grader bias and tutor disagreement
- Noisy and incomplete data
- Designing automated feedback that is fair, constructive, and useful
- Building evaluation pipelines you can actually trust
A practical look at architecture, evaluation, and feedback design for teams building reliable, interactive AI systems at scale.
About the AI Lightning Round
Fast. Focused. Practical. Short talks from builders deploying AI where it actually runs — part of the Kiss the Future Meetup, organized by Bloomteq in partnership with SLO AI.