Projects

Home Research Publications Projects Service Updates Blog CV

Selected code and research prototypes aligned with my two main directions: GenAI optimization and AI hardware security.

GenAI Optimization

TOGGLE

Temporal-logic-guided compression for edge LLM deployment, focused on balancing efficiency with behavioral constraints.

FairCompress

Research prototype for compression methods that account for efficiency, fairness, and deployability rather than size alone.

VeriNAIS

Verified architecture search for AI systems where deployment behavior must satisfy formal constraints.

OSPA Transformer

Attention mechanism work aimed at improving transformer efficiency through structured subspace projections.

AI Hardware Security and Robustness

flipRL

Prototype connected to efficient bit-flip attacks and accelerator-level vulnerability assessment for LLM systems.

VERMITHOR

Formal runtime control for edge cyber-physical inference where thermal and reliability constraints are first-class requirements.

Approximate Computing

Low-level experimentation around approximate computation, fault behavior, and the security-reliability tradeoffs of efficient AI.

Neural-Pulse

Signal Temporal Logic monitoring for hallucination-like behavior and semantic attack patterns in LLMs.

Model Understanding

GhostTrack

Framework for tracing competing thought trajectories and identifying hallucination risks before final output.

Attention Sink Analysis

Tools for analyzing first-token dominance and attention sink behavior in long-context language models.