Cognitive * Corpus
I build AI systems that turn complex documents and connected knowledge into persistent, traceable memory—giving agents the grounded context they need to do reliable work.
About
I am co-founder of Hashtag.AI, where I now work full-time on reliable knowledge infrastructure for AI agents. We turn large corpora of complex documents into accurate, traceable context that agents can use across real workflows.
I design end-to-end AI workflows and agent harnesses: document ingestion, OCR and structured extraction; retrieval, tools and orchestration; evaluation, guardrails, observability and deployment. The stack spans FastAPI, embeddings, vector databases, Neo4j, LangChain, LangGraph, GraphQL and GCP.
My current focus is cognitive memory: systems that continually integrate new knowledge and recover connected evidence from partial cues. Drawing on research into hippocampal memory indexing, I explore pattern separation during encoding, pattern completion during retrieval, OpenIE knowledge graphs, retrieval encoders and graph-based ranking for associative, multi-hop recall.
Previously, I was a Senior Generative AI Engineer at Quantaco Technology and an AI/ML research fellow. My PhD in Mathematical Physics still shapes how I work: model the structure, reason across relationships, and make every system testable and auditable.
Work
Updates
Field notes from my work in AI, cognitive memory, knowledge systems and research.