Skip to content
View jmiguelmangas's full-sized avatar
:fishsticks:
Working Smart
:fishsticks:
Working Smart

Block or report jmiguelmangas

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. 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
jmiguelmangas/README.md

Hi, I'm Jose Miguel Mangas

AutoPulse AeroRoute MLX Python React MongoDB Docker OpenAPI

I'm a product-minded Python / full-stack developer based in Madrid, working at the intersection of business analysis, operational tooling, automation and data products.

My work is focused on understanding business workflows, turning messy operational needs into usable tools, and building systems that make decisions, reporting and daily execution easier. Outside work, I build portfolio projects that combine backend engineering, data ingestion, geospatial search, analytics, automation, synthetic trajectory modeling and polished frontend experiences.

What I Build

  • Internal tools for operations, planning, reporting and workflow automation.
  • Data products with APIs, ingestion pipelines, dashboards and quality checks.
  • Full-stack applications that connect business context with clean technical execution.
  • AI-assisted and automation-first developer workflows.

Featured Projects

AutoPulse

Automotive intelligence platform for EV charging infrastructure.

  • FastAPI backend with station search, route planning, analytics and admin data quality APIs.
  • MongoDB geospatial data model with Open Charge Map ingestion.
  • React + TypeScript frontend with MapLibre, bento-style UI, station cards, modeled availability and route recommendations.
  • Shared OpenAPI/TypeScript contracts, CI, metrics and structured logging.

Repos:

AeroRoute MLX

Educational flight-trajectory efficiency simulator: a deterministic 4D route optimizer, a PostGIS-backed FastAPI service, a React/MapLibre frontend, and a locally-run Apple Silicon LLM (MLX, Gemma 3) that explains — but is contractually forbidden from computing — the result.

AeroRoute MLX: route comparison, deterministic AI explanation and operational-safety gates

  • Deterministic core, probabilistic explanation, strictly separated. A custom layered label-setting solver computes routes, fuel and scoring; the local LLM only narrates an already-computed result, with output validators (not just prompting) that reject any invented or mismatched number.
  • A measured model bake-off, not a preference call. Chose the base model by running a 24-case evaluation corpus against 3 candidates and recording pass rate, latency and memory — including a case where a challenger failed most cases, and documenting why rather than discarding the result.
  • Fail-closed safety, visible at runtime. Filing/dispatch/operator- approval gates are real API fields with automated regression tests that fail if a future endpoint quietly weakens the non-operational boundary.
  • Eight independently versioned repositories (web, API, optimizer, native MLX service, offline training, data curation, cross-language contracts, platform) with a pinned release manifest and a guardrail that fails the build if a component's pinned version drifts from what's actually checked out.

Repos:

Operations and Productivity Tools

Personal projects around crew/operations workflows, internal productivity, automation, dashboards and business process support.

Examples from my workspace include Crewfinder, Crewcast, BambooHR time tracking automation and operational pathfinder tools.

Tech I Use

Backend and data: Python, FastAPI, Django, Pydantic, SQL, MongoDB, PyMongo, HTTPX, data validation, API integrations.

Frontend: React, TypeScript, Vite, TanStack Query, MapLibre, HTML, CSS, design tokens and reusable component systems.

Automation and operations: CLI tools, scheduled jobs, ingestion workers, business process automation, reporting workflows.

Quality and delivery: Pytest, Ruff, Vitest, ESLint, OpenAPI, generated contracts, GitHub Actions, Docker, observability and structured logging.

AI workflow: Codex, prompt-driven development, rapid prototyping and AI-assisted architecture exploration.

Current Focus

  • Building portfolio-grade products that show both engineering depth and business understanding.
  • Turning operational workflows into reliable software.
  • Improving frontend/product design so tools feel clear, useful and polished.
  • Connecting APIs, geospatial data, analytics and automation into cohesive applications.

Contact

Pinned Loading

  1. SEJO_SDK SEJO_SDK Public

    Lightweight Python SDK for provider-agnostic AI agents across OpenAI, Anthropic, Gemini and DeepSeek.

    Python 1

  2. styleagent-platform styleagent-platform Public

    Python 1

  3. AutoPulse AutoPulse Public

    Portfolio-grade automotive intelligence platform for EV charging, routing and data quality.

  4. aeroroute-platform aeroroute-platform Public

    AeroRoute MLX architecture, local development, integration testing, observability and release manifests.

    Python