Jose R.
Founder & Builder
I build AI-powered tools that solve real everyday problems — from job hunting to personal productivity and beyond.
What I've Built
Live applications and the infrastructure powering them.
JobSentry
LiveMulti-service scam-detection platform for job postings. A React frontend routes through a Node.js gateway to two specialized backends: a Quarkus/Java service for WHOIS domain-age lookups and a FastAPI service that streams LLM analysis via a local Ollama instance on a Jetson Orin Nano. Flags suspicious postings in seconds using domain age, linguistic red flags, and recruiter patterns.
Wellness Portal
LivePersonal wellness tracking app where users log symptoms, moods, sleep, and daily habits. A FastAPI backend aggregates the entries and sends them to a locally-hosted LLM (Ollama on Jetson Orin Nano) to generate personalized pattern insights — no data ever leaves the home lab.
StockFinancia
LivePersonal finance dashboard with a composite PulseScore (1–10) built from RSI, MACD, EMA crossovers, Bollinger Bands, options flow, insider activity, and earnings proximity. Tracks watchlist stocks with intraday snapshots, 7/30/90-day verdict accuracy tracking, a paper trading ledger, AI-powered chat analysis, and real-time WebSocket announcements. Fully self-hosted on Kubernetes.
The Stack Behind It
Everything runs self-hosted on bare metal — no cloud bills.
K3s Home Lab
Self-hosted 5-node Raspberry Pi cluster (1 control plane + 4 workers) running Kubernetes via k3s. Hosts 15+ production services: ArgoCD for GitOps deployments, Longhorn for distributed block storage, MinIO for S3-compatible object storage, Grafana + Prometheus for observability, Umami for privacy-first analytics, and a private Docker registry. All nodes run ARM64 with persistent storage across reboots.
CI/CD Pipeline
Jenkins shared-library pipeline that handles all projects from a single Jenkinsfile template. Runs parallel build + test stages using Docker BuildKit with layer caching, automatically bumps semantic image tags in kustomization.yaml, reports commit statuses back to GitHub, and triggers ArgoCD syncs. SCM polling every 10 minutes with webhook fallback.
AI Inference Node
NVIDIA Jetson Orin Nano (8 GB, 1024-core Ampere GPU) configured as a dedicated edge inference node. Runs Ollama serving llama3.2:3b with GPU acceleration via the NVIDIA container runtime. Receives inference requests from JobSentry's linguistic service and Wellness Portal over the cluster's internal network — fully air-gapped LLM inference with no external API costs.
Observability Stack
Grafana + Prometheus deployed via Helm across all cluster nodes. Custom dashboards track CPU/memory per node, container restart rates, Longhorn volume health, and application-level metrics. Alertmanager routes critical alerts. Umami provides privacy-first web analytics for all public-facing apps without third-party data sharing.
Podoi Tech LLC
I'm Jose R., an indie developer and founder of Podoi Tech LLC — a one-person studio focused on building practical AI products.
My work sits at the intersection of artificial intelligence and everyday usefulness. I believe the most impactful software solves problems that real people face every day — not just enterprise edge cases.
I run a self-hosted infrastructure stack on Kubernetes (k3s), ship continuously with ArgoCD, and obsess over clean, maintainable code.
Tech I work with
Get in Touch
Have an idea, question, or just want to say hi? I'd love to hear from you.
© 2026 Podoi Tech LLC. All rights reserved.