← All work

Industry work

AI Evaluation Tool

Assessment, verification and correction workflows for generated reports.

2024–2025Role period

The work

Deployed an AI application on NVIDIA DGX and Kubernetes for a $16M engagement. Automated LLM report assessment, verification and correction; factual accuracy improved from 82% to 91% in the reported evaluation.

Context

Factuality assessment, retrieval and enterprise AI infrastructure.

Scope

Data Scientist at Deloitte; engineering and deployment of automated LLM report assessment, verification and correction.

Confidentiality

Client source code and implementation details are not publicly shared.