knowledge/tools.md

# Lab Tools & Systems

## ResearchAtlas — graph-based literature discovery
A system that helps researchers map a field rather than just rank hits. Instead of
returning a ranked list like a standard search engine, ResearchAtlas is
**coverage-oriented**: it builds a graph over the literature so a user can see the
structure of a research area, discover related work, and navigate connections
between papers and ideas. It reflects the lab's broader thesis that **graph structure
is a first-class signal** for understanding and retrieval.
Link: https://www.researchatlas.dpdns.org/

## LLMCostCut — cutting LLM API cost by up to 10×
A method/toolkit that reduces the number of expensive large-LLM API calls **by up to
10×** through **selective invocation** (only calling the large model when it is
actually needed) combined with **online distillation** (learning from the large
model's outputs so a cheaper model can handle more cases over time). This is the
research basis for the **cost-aware model routing** used inside this very assistant:
cheap questions are answered by a small/fast model and only substantive research
questions escalate to a stronger model.
Link: https://github.com/zhaoliangvaio/llmcostcut

## How this assistant embodies the lab's research
This lab assistant is itself a small demonstration of several of the lab's lines of
work:
- **Retrieval-augmented generation (RAG)** over the lab's publications — see *GRAG*
  and *Spatial-RAG*.
- **Citation-/graph-aware retrieval**: after vector search, results are expanded one
  hop along a paper graph (shared authors and topics), echoing *GRAG* and *CG-RAG*.
- **Trustworthy generation**: answers are grounded in retrieved sources with inline
  citations, and the assistant declines to answer when the corpus does not support a
  claim — reflecting the lab's emphasis on robustness and uncertainty.
- **Cost-aware routing**: a direct nod to *LLMCostCut*.

## Note on "Causal Dynamics Lab" / Cielara
Separately, Dr. Zhao is involved as a research scientist with the **Causal Dynamics
Lab**, an AI lab whose product **Cielara** builds a "production world model" to
simulate code changes against live environments before deployment. This is an
industry effort distinct from his Emory academic group; for academic research,
advising, and publications, refer to the Emory lab.