Projects
A few AI, software, and research projects I’ve contributed to. Professional work is summarized at a high level.
Reliability for multi-agent diagnosis
Developed grounding guardrails and a multi-judge evaluation methodology for root-cause analysis agents. Focused on catching unsupported conclusions and measuring diagnosis quality against expert-resolved references.
Making retail data conversational
Developed an agentic retrieval assistant for natural-language data access. Paired retrieval and answer evaluation with ML data pipelines and APIs for downstream consumers.
ML inference for fleet routing
Built Python inference services and integrated routing models with operational software. Emphasized availability, fast response times, and scalable deployment.
FACTIFY3M
Contributed to benchmark development and evaluation for multimodal fact verification, bringing text, images, paraphrases, adversarial stories, and explainability into one research setting.