Back home
JarnDev

José Alfredo Ribeiro Neto

Backend & Full-Stack Engineer

Bahia, Brazil

Backend & full-stack engineer on TypeScript/Node and Python, building for North-American startups. From computer-vision detection and LLM & graph pipelines to a production IoT fleet and payment & FX ledgers — I rebuild production systems without switching them off, and ship things that stay up and stay observable.

alfredo+portifolio@jarndev.com.brGitHubLinkedIn
01

Experience

Keep

Backend Engineer

Fintech · Toronto, CA

B2B fintech serving ~6,000 Canadian businesses — corporate cards, multi-currency banking, FX and capital. I worked on the money-movement core, where a race condition is a financial loss, not a stack trace.

  • Project lead on seven cross-functional banking & infrastructure initiatives — RFC through phased rollout to retiring the legacy path
  • Redesigned the bank-payment lifecycle and migrated inbound funding to an event-driven pipeline with zero downtime
  • Eliminated a recurring reconciliation incident by moving wallet balances to the FX provider as source of truth — pending-aware validation and a double-deduction guard
  • Incident lead on ~10 production incidents; resolved ~45 customer & internal escalations — stuck payments, balance drift, failed settlements, SWIFT pacs.008/UETR wire traces
  • Persisted a record for every transaction-monitoring outcome — not just approvals — then removed the legacy path entirely
  • Hardened the platform: moved nine vendors' credentials into AWS SSM/Secrets Manager and added HMAC verification to vendor webhooks
  • Authored six architecture RFCs and built the Datadog dashboards, alerts and Metabase reporting for the systems I owned
  • TypeScript
  • Node.js
  • GraphQL
  • Prisma
  • PostgreSQL
  • Redis
  • RabbitMQ
  • AWS
  • Datadog

Axis Worldwide Inc

Full-Stack Engineer

New York, USA

A strategic-intelligence platform that aggregates global events, regulations and actors so companies and governments can act on them.

  • Built and integrated LLM pipelines over a feed of ~2M events/day for entity and event extraction
  • Shipped the first graph-network viewer over ~100k nodes and millions of edges; migrated the legacy dataset to Neo4j
  • Sped up core reads by profiling hot queries and reworking the schema; wired external APIs to ingest and serve data
  • React
  • Python
  • OpenAI
  • Embeddings
  • Node.js
  • Neo4j
  • PostgreSQL
  • Redis
  • Kubernetes

Hugo → Calendly

Backend Engineer

Hugo acquired by Calendly · California, USA

Hugo consolidates a meeting's key features — notes, tasks and 20+ integrations — in one place; the team was acquired by Calendly, where I kept building.

  • Rewrote a core algorithm around tokenizers, improving its performance 500×
  • Fixed a memory leak that was crashing the service every 12–24h — no further restarts after the fix
  • After Calendly acquired the team: improved the internal CLI and Slack-integrated CI/CD bot for release automation
  • Built Yjs-based real-time collaboration and tuned AWS auto-scaling via CloudFormation
  • Node.js
  • React
  • TypeScript
  • Yjs
  • AWS
  • CloudFormation
  • Ruby on Rails

Suflex

Full-Stack Engineer

FoodCamp · Salvador, Brazil

Startup helping food companies thrive with organization and process control.

  • Migrated the FoodCamp backend from Elixir + REST to NestJS + GraphQL
  • Owned data modeling, tech choices and hosting; migrated infra to Kubernetes on AWS
  • Built and ran a production IoT fleet of ~3,000 label-printer devices in restaurant kitchens
  • On a 5-person team — bi-weekly architecture reviews, knowledge-sharing and interviewing candidates
  • Designed the GraphQL schema and data model behind the platform
  • Ran over-the-air updates for the printer fleet — remote deploy and recovery across live kitchens
  • Node.js
  • NestJS
  • GraphQL
  • Kubernetes
  • AWS

AmbevTech

Python Backend Developer

Salvador, Brazil

Ambev's technology arm, building internal products for the brewing giant.

  • Built the Python boilerplate adopted by 4+ projects as the standard — conventions, patterns and test setup
  • Designed the FrostSignature data model and its Python + FastAPI backend, with 75%+ test coverage throughout
  • Python
  • FastAPI
  • MongoDB Atlas
  • PostgreSQL
  • Docker
  • Azure

Aton Engenharias

Software Developer

Salvador, Brazil

Telecom-engineering and surveillance company (est. 1988). Built forest-fire detection across 200+ monitored cameras with deep learning and computer vision.

  • Raised detection-model accuracy from 25% to over 80%
  • Managed the internal web infrastructure across 11 servers between client and development sites
  • Built the detection pipeline end-to-end — data collection, model training and inference on live camera feeds (OpenCV, TensorFlow/Keras)
  • Python
  • C / C++
  • OpenCV
  • TensorFlow
  • Keras

Earlier roles

  • QA Analyst · BRISA · LG MobileSalvador, Brazil
  • Assistant Researcher · University of Illinois at ChicagoChicago, USA
  • IT Technician · Wyden Area1Salvador, Brazil
02

Technical skills

Core
TypeScriptNode.jsGraphQLPostgreSQLAWSKubernetes
Backend
NestJSFastAPIRESTRabbitMQEvent-driven architecture
Languages
JavaScriptPythonC / C++SQL
Data
MongoDBRedisNeo4jPrisma
Cloud / DevOps
DockerCloudFormationDatadogSentry
AI / ML
LLMsOpenAIEmbeddingsComputer VisionDeep Learning
Robotics & Embedded
C / C++ArduinoIoTROSGAZEBOVREPRVIZ
03

Education

  • B.Sc. in Mechatronics Engineering (suspended)

    Wyden Area 1

    2011

  • Exchange — Electrical Engineering (Mechatronics concentration)

    Tennessee Tech University

    2015 — 2016

04

Certificates

  • CAD 2D · Wyden Area 12012
  • CTFL Foundation · BSTQB2017
  • C Programming for Embedded Applications · LinkedIn2020
  • Scrum Foundation Professional Certificate · CertiProf2020
  • DataScience Acceleration · Codenation2020
05

Languages

  • English · Advanced
  • Portuguese · Native
  • Spanish · Basic
06

Projects

Curated from GitHub — grouped by the thread that ties them together.

Vision & AI

Mechatronics is where I started, and perception keeps pulling me back — from computer vision to orchestrating local AI models.

  • ai-cli

    A unified CLI + TUI that orchestrates local image, video, voice, music and LLM models from one command, on a single GPU.

    • Python
    • Typer
    • Textual
    • LLMs
    • Diffusion
  • OurMall-Vision

    Hackathon project: a computer-vision filtering system integrated with a mobile app.

    • React Native
    • Computer Vision
    • Python

Products, shipped

Full applications taken from data model to a deployed, usable UI.

  • powerup

    Production-ready multi-tenant monorepo — a Next.js App Router front end and an Express + PostgreSQL API, with shared UI, types and validation packages.

    • Next.js
    • Express
    • PostgreSQL
    • Monorepo
  • Farol Eleitoral

    Data app that turns Brazilian electoral datasets into a browsable, filterable dashboard.

    • Next.js
    • shadcn/ui
    • TypeScript
  • force-graph-lab

    Interactive force-directed graph explorer built on d3-force for probing relationships and layouts.

    • React
    • d3-force
    • Vite

Client work

Production sites shipped for real businesses through JarnDev — Astro + Tailwind, fast and static.

  • Lebre Tecnologia

    Institutional site for an IT & technology services company.

    • Astro
    • Tailwind CSS
  • Oficina do Clóvis

    Site for a Salvador auto shop — services, 24h towing and vehicle insurance.

    • Astro
    • Tailwind CSS

Backend & tooling

Small, sharp services and the plumbing that keeps things running.

  • flowdeck

    Multi-repo gitflow visualizer — shows what's already in each stage, with cherry-pick detection by patch-id (not just SHA).

    • Node.js
    • Git
    • TUI
  • osint-docker

    Isolated, container-per-tool OSINT environment — each tool runs sandboxed, with no access to the host.

    • Docker
    • OSINT
    • Security