Roptal — BYOC ML deployment platform

Gallery

About

Roptal is a working bring-your-own-cloud (BYOC) deployment platform for ML teams. Connect a GitHub repo; Roptal scans the runtime (Python version, CUDA, ports, system libs, model files), generates a reviewable Dockerfile from templates for 10 ML frameworks, builds inside the customer's own cloud account, deploys, and then operates it: live SSE logs, health checks, cost explorer, LLM-powered log analysis, HMAC-signed webhooks, canary / blue-green traffic splitting and rollback.WHAT WORKS IN PRODUCTION TODAYAWS: SageMaker real-time and serverless endpoints, CodeBuild + ECR image builds, API Gateway, autoscaling, IAM auto-setup, CloudWatch logs, GPU instances, rollbackAzure: Container Apps + ACR builds, canary and blue/green traffic splitting (CPU only)Hugging Face Spaces with GPU tiersModel importers: GitLab, Hugging Face, S3, Torch Hub, notebooksConsole with 19 pages: deployments, monitoring, models, billing, workspaces, audit log, API keys, adminCLI on PyPI (pip install roptal-cli): local env scan, Dockerfile generation, deploy, status, logsMCP server (scan_repository, deploy_model) usable from Claude Code, Cursor and other agentsAuth: Supabase Auth, TOTP 2FA, hashed API keys, GitHub App with installation ownership checksBilling: Paddle, credit-based with atomic holds, reversals and ledger historyMarketing site, blog with auto cross-posting to dev.to / Hashnode, docs site, full legal stack (ToS, Privacy, DPA, Subprocessors, AUP, Refund, AI Disclosure)SCAFFOLDED, NOT LIVE (buyer's roadmap)GCP Cloud Run, RunPod, Railway, Render, Vercel and Oracle deployer stubsCross-cloud migration (built, disabled in production)TECH STACKPython 3.11 / FastAPI / Celery / Pydantic backend with a small Go reverse-proxy gateway. React 18 + TypeScript + Vite + Tailwind frontend. Supabase Postgres with row-level security, Redis, Sentry, Resend, OpenAI (Anthropic / Groq as alternatives). Backend runs on Azure Container Apps via GitHub Actions; frontend and docs on Vercel.SIZE AND QUALITY~105k lines of code (58k backend, 46k frontend), ~15k of it tests. 1,358 pytest tests, 241 Vitest tests, 15 Playwright e2e tests, 11 Go tests, all run in CI on every PR and before every production deploy. Built Dec 2025 - Oct 2026, ~900 commits, one primary author. Clean repos: no secrets on main, proprietary LICENSE in place, dependency licenses audited (no GPL/AGPL at runtime).INCLUDED IN THE SALEBoth source repos (backend + frontend/console/docs), roptal.com + docs.roptal.com + api.roptal.com, the roptal-app GitHub App, the roptal-cli PyPI package, the LinkedIn page, the narrated walkthrough video, all blog content, and an IP assignment.WHY I'M SELLINGCoding agents plus cloud CLIs have commoditized the "repo to endpoint" step. The operating plane is still valuable, but it needs an owner with existing ML or DevOps clients to sell it into. A consultancy can white-label this on day one; a solo founder with no customer base can't.TRACTIONPre-revenue, zero users, never ran ads. This is a code and asset sale, priced far below rebuild cost. Running costs are Azure, Supabase, Vercel and OpenAI usage; near zero while idle.IDEAL BUYERAn MLOps / DevOps consultancy, an Azure-focused shop, or a small ML platform company that wants a finished deployment product instead of building one.Price is fixed at $5,000 USD, paid via escrow. Happy to walk through the code on a call and give read access for diligence under NDA.