Skip to content

v1.0 - Entangle Core: GitHub Quantum Ecosystem Engine

Choose a tag to compare

View all tags
@aangell98 aangell98 released this 07 May 18:53
· 40 commits to main since this release
v1.0
95e4315
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

🌌 Entangle Core v1.0

First public release of the analytics engine behind Entangle: a research‑grade platform that maps, analyses and visualises the open‑source quantum computing ecosystem on GitHub.

Built as the backend of a Bachelor's Final Project (TFG) at the University of Castilla‑La Mancha. From an empty MongoDB to a live, AI‑augmented dashboard running on Azure, end to end.

✨ Highlights

🔭 Six‑stage ingestion pipeline

  • Crawls GitHub for 70+ curated quantum keywords (Qiskit, Cirq, PennyLane, Braket, OpenQASM, QML, NISQ, …) bypassing the 1.000‑result Search API limit through star/year segmentation.

  • Combines GraphQL super‑queries with parallel REST calls (ThreadPoolExecutor, 4–7 workers) for repos, users and organisations.

  • Heuristic filters discard false positives (Firefox Quantum, CSS libraries, …) and the pipeline is incremental: only new or modified entities are re‑fetched.

  • Computes derived signals: quantum_expertise_score for users, quantum_focus_score for organisations and discipline classification across five technical domains.

🕸️ Network analysis

  • Builds a heterogeneous graph (~28k nodes, ~98k edges) with NetworkX and detects 752 communities with the Louvain algorithm.

  • Identifies bridge users (multi‑organisation contributors) and discipline bridges (cross‑domain contributors).

  • Computes betweenness/degree centrality with adaptive sampling (k=50 over 5k+ nodes) and a bus‑factor indicator for resilience.

🤖 Conversational AI agent

  • Router–Worker architecture powered by GPT‑4o: a deterministic router classifies the user's intent (DATA, DASHBOARD, UNIVERSE) and dispatches it to a specialised worker.

  • Server‑Sent Events (SSE) streaming for token‑by‑token answers.

  • Each worker exposes scoped MongoDB tools: no arbitrary aggregations, no injection surface.

🚀 Production‑ready API

  • FastAPI with async I/O, GZip compression (~87% saving on the 22 MB collaboration graph) and a three‑level cache (in‑memory → MongoDB → recompute).

  • Sub‑100 ms latency on hot endpoints; full network metrics in ~1 s.

  • orjson for ~10× faster JSON serialisation.

☁️ Cloud‑native deployment

  • Azure Container Apps + Cosmos DB vCore M30 (the migration from RU to vCore boosted ingestion 15–20×).

  • Bicep Infrastructure‑as‑Code under infra/.

  • GitHub Actions workflows for production and staging.

  • Continuous quality enforcement via SonarCloud Sonar Way gate (passing on every PR).

📊 By the numbers

Metric Value
Commits (Oct 2025 → Apr 2026) 160
Curated quantum keywords 71
Graph users ~27.000
Graph edges ~98.000
Communities (Louvain) 752
Hot‑endpoint latency <100 ms
Test suite pytest + pytest‑cov, ~61% line coverage
Quality gate ✅ SonarCloud Sonar Way, passing

🛠️ Tech stack

Python 3.11 · FastAPI · Pydantic · MongoDB (Azure Cosmos DB vCore) · NetworkX · OpenAI GPT‑4o · Docker · Bicep · GitHub Actions · SonarCloud

🔗 Companion projects

🙏 Acknowledgements

Designed, built and maintained by Ángel Luis Lara Martín under academic supervision at UCLM.

Released under the MIT License, see LICENSE.

Assets 2
Loading