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 …
Building the future, one commit at a time. Here are some of my open-source projects and research work.
A false association does not have to be learned during pretraining. It can be preserved and amplified by the retrieval …
When should a streaming pipeline pay for an LLM call? Cast as a risk-based sequential stopping problem, with six proved…
Accepted at C3NLP 2026 @ ACL. The first open-source system for bidirectional interaction with cuneiform — read Akkadian…
A Genshin Impact-inspired Windows CPU optimiser that keeps Intel P-cores active via adaptive multi-threaded load, deliv…
A Rustlings-style interactive exercise system covering 63 exercises across 6 phases — from Euclidean review through hyp…
A CUDA-level profiling and intelligent multi-model inference scheduling framework for edge AI, achieving 114.67 FPS YOL…
An interactive Rustlings-style platform for implementing every reinforcement learning algorithm from Sutton & Barto…
Browser-native vector database in Rust compiled to WebAssembly — HNSW + Flat indices, SIMD128 acceleration for 7 distan…
Multi-paradigm fractal renderer spanning 5 compute backends (JS → WASM → CPU → OpenMP → CUDA) with 125,000x GPU speedup…