Polyglot Grounded QA
Polyglot Grounded QA
Multilingual grounded QA pipeline focused on citation-aware answering and abstention, with artifact-backed evaluation instead of demo-first output.
Projects
Tool Calling Reliability Benchmark
Tool Calling Reliability Benchmark
Benchmark harness for testing LLM tool-calling under retries, stalls, timeouts, and failure-transfer scenarios instead of ideal-path demos.
midi-mixer (Go TUI)
A terminal-based MIDI mixer built in Go. Keyboard-driven workflow for mixing and routing—built as a focused developer tool.
OCN Agents (OneCarNow)
OCN Agents (OneCarNow)
Production agentic platform for chat, voice, and KYC workflows across Django services, async workers, retrieval and tool execution, and React delivery.
TurboQuant x Refusal Eval Bench
TurboQuant x Refusal Eval Bench
KV-cache compression benchmark that measures how TurboQuant-style caching shifts refusal behavior, fidelity, and long-context robustness under matched prompts.
Go Rig
Go Rig
Bare-bones Go-based coding agent framework built from scratch. Inspired by the Pi architecture: small core abstractions, append-only JSONL sessions, tool execution as events, and multiple user-facing modes on one runtime.
RLM Harness
RLM Harness
Go TUI that turns any folder into a knowledge base via Recursive Language Model methodology. Implements Bubble Tea TUI, Cobra CLI, pluggable LLM providers with live trajectory visualisation.
Refusal Suppression
Refusal Suppression
Activation-geometry pipeline to locate, ablate, and evaluate refusal direction in Llama-3.1-8B-Instruct. Found refusal behavior geometrically localised to single direction at layer 14; ablation suppressed refusal from 0.98 to 0.16.
