Skip to content
View byQuexo's full-sized avatar
👽
👽

Organizations

@iLert

Block or report byQuexo

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
byQuexo/README.md

AI/LLM Engineer @ ilert

Production LLM agents for alerting and incident response at ilert, Köln.
Multi-agent / A2A · MCP · realtime voice · Rust LLM gateway · AWS Bedrock · evals and guardrails

Live demo docrag-bench  npm agent-approval-gate  crates.io agent-approval-gate


Projects

Project What it does Result
Warden
Rust · TypeScript (WASM)
Authorization for AI agents: every tool call checked against the user, the agent chain and the exact arguments, outside the model. story 216 of 216 scenario and generated attack cases pass, 0 violations
attack and scenario cases, 0 violations
Steward
Rust · React
Company assistant on Warden and Rig: nothing is written in your name without your confirmation, answers cite verified sources. story 20 of 20 demo conversations pass end to end in CI
demo conversations end to end in CI
docrag-bench
Python
PDF question answering with verified citations, plus a benchmark of what actually helps. 0 → 0.83 recall@10 on scanned PDFs
recall@10 on scanned PDFs
agent-approval-gate
TypeScript · Rust
Human-in-the-loop approval for LLM agent tool calls. On npm and crates.io. 51 tests, 42 TS + 9 Rust
tests, 42 TS + 9 Rust
Alert-Grouping-Evaluation
Python
BSc dissertation: grouping Prometheus alerts with multilingual embeddings. 0.91–0.95 F1 on EN / FR / ES
F1 on EN / FR / ES
incident-triage-demo
TypeScript
On-call SRE agent on Gemini that asks before it acts. 3 risk tiers, reads run, writes wait
risk tiers, reads run, writes wait

Upstream: Rig

Rig is the Rust LLM framework behind ilert AI's agentic LLM proxy. Three merged fixes · all PRs

PR Change Merged
#1725 fix(mistral): expose cached and audio token fields in Usage 4 May 2026
#1683 fix(bedrock): preserve adaptive-thinking signatures in streaming reasoning 28 Apr 2026
#1675 fix(bedrock): handle adaptive-thinking interactions for prompt caching and reasoning conversion 28 Apr 2026

Stack

Rust, TypeScript, Python, AWS Bedrock, Qdrant, Hugging Face, MCP


contribution snake

Pinned Loading

  1. docrag-bench docrag-bench Public

    Question answering over real PDFs with hybrid retrieval, reranking and verified citations, plus a benchmark that measures which choices actually help (0.544 correct on MMLongBench-Doc, with confide…

    Python

  2. agent-approval-gate agent-approval-gate Public

    Human-in-the-loop approval gate for LLM agent tool calls

    TypeScript 1 1

  3. iLert/mcp-ilert iLert/mcp-ilert Public

    MCP server for ilert

    17 1

  4. 0xPlaygrounds/rig 0xPlaygrounds/rig Public

    ⚙️🦀 Build modular and scalable LLM Applications in Rust

    Rust 9.2k 1k