< Projects />

COMPLETED

Research

Index-Level Adversarial Memorization — When Retrieval Encodes Wrong Entity-Context Bindings

A false association does not have to be learned during pretraining. It can be preserved and amplified by the retrieval …

Python BM25 BGE-M3 mE5 +3

EXPLORE()

PUBLISHED COMPLETED

Research

Event-Triggered LLM Invocation — Uncertainty-Aware Sequential Decision Rules

When should a streaming pipeline pay for an LLM call? Cast as a risk-based sequential stopping problem, with six proved…

Python PyTorch vLLM CMAPSS +3

EXPLORE()

PRIVATE ACTIVE

Research

IR-KG-LLM-Alignment

Historical narrative conflict detection using BM25 + knowledge graphs + belief graph reasoner, in preparation for ACL R…

Python Knowledge Graphs BM25 LLMs

EXPLORE()

PRIVATE ACTIVE

Research

EasyWorldModel

Uncertainty-calibrated residual world models with 18% lower prediction error and deep ensemble uncertainty decompositio…

Python PyTorch Hydra

EXPLORE()

PRIVATE ACTIVE

Research

EdgeRoutingICML

Guardian-Analyst two-tier edge intelligence with sub-2ms HMM routing on Jetson Orin Nano.

Python HMM TinyLlama

EXPLORE()

PUBLISHED ACTIVE

Research

Interruptr

Multi-agent C/C++ vulnerability detection achieving ~96% detection rate vs Cppcheck 12%, with 95% cost reduction vs GPT…

Python AutoGen LangGraph

EXPLORE()

PRIVATE ACTIVE

Research

CrystleLLM

Quotient space reasoning for LLMs: identifies 64-72% semantic redundancy and reduces token usage by 26.8%.

Python vLLM GSM8K MMLU +4

EXPLORE()

PUBLISHED ACTIVE

Research

ROA-LLM

A systematic security study formalizing Persistent Semantic Entities (PSEs) as a new attack surface in tool-augmented L…

Rust (PyO3) Python 15 LLMs from 7 vendors Factorial Ablation Design +1

EXPLORE()

PUBLISHED ACTIVE

Research

EdgeCloud-Reasoning

Mechanism-designed belief aggregation for edge AI: VCG-based truthful elicitation achieves lower false negative rates t…

Python Mechanism Design DeepSets Transformer

EXPLORE()