Industrial AI & MiningOps
MiningOps Copilot — Local Mining POC
A locally installed POC that adapts ASIC, energy, and environment CSV files, detects deterministic anomalies, and asks Qwen3 for explanations.
- Input
- CSV / JSONL
- Polling window
- 10-30s
- when the source is available
Challenge
Adapt heterogeneous operational files and produce understandable incidents without automatically controlling equipment.
Solution
FastAPI polls CSV/JSONL files, applies deterministic rules, records incidents, and uses Qwen3 through Ollama for explanations and suggested actions.
Available evidence
- Docker/FastAPI stack and UI installed in the client environment
- Deterministic anomaly detection
- JSONL incidents and daily summary
- Source
- Integracao_LLM_Mineracao@27fb5b0 + local installation report dated 2026-04-06 (unversioned)
- Period
- January-February 2026 data; March commit; installation reported on 2026-04-06.
- Method
- Pipeline, code, status, and local installation-report review; no connection to the client environment.
- Context
- File- and polling-based local-first POC; scope is analysis, not automatic ASIC control.
- Verified on
Limitations
- The later installation report is unversioned.
- The correct dataset and automatic ingestion had not received full acceptance in the latest evidence.
Architecture & Stack
The stack reflects the recorded implementation and this project's context; it is not a generic guarantee of resilience, performance, or scale.
- Python
- FastAPI
- Ollama
- Qwen3
- Docker
- CSV
- JSONL
Core Features
- CSV adaptation
- Rule-based detection
- JSONL incidents
- Daily summary
The system on screen
Real interface with demonstration data. Does not represent customer outcomes.