Two engineers reviewing an operational web application across dual laptops

About Oruvyn

Direct engineering for US operators.

You work straight with Caleb Njiiri — an engineer who reads your ERP tables, quote sheets, and dispatch boards before writing a single line of code.

Get a free workflow snapshot

Why Oruvyn exists

Small and mid-sized companies between 15 and 250 employees sit in an underserved gap. Enterprise consultancies pitch six-figure discovery phases staffed by junior analysts, while low-code agencies string together fragile webhooks that break the first time a part number is discontinued or a carrier changes a billing code.

Caleb Njiiri started Oruvyn to give US operators a direct principal-engineer alternative: someone who understands margin math, sits with your counter reps or dispatchers on a busy Tuesday, and ships clean production software tied straight into the ERP, practice management, or field service tools you already pay for.

Who we work with best

Specialty wholesale distributors, residential and light-commercial HVAC/plumbing/electrical contractors, multi-site dental and medical groups, custom fabrication job shops, and B2B ecommerce teams where quoting, scheduling, or order intake still depends on spreadsheets and one overburdened veteran employee.

Caleb Njiiri, founder and principal engineer at Oruvyn
Caleb Njiiri · Founder & Principal Engineer · [email protected]

Commercial & ownership commitments

  • 100% client-owned IP: Upon milestone payment, your company owns the Git repository, database schemas, deployment configs, and prompt templates outright. Zero seat fees or hostage licensing.
  • Fixed-scope milestones: Every build starts with a written specification, explicit acceptance criteria, and a fixed milestone price — typically $6.5k–$15k for a 30-day pilot queue and $18k–$45k+ for full internal workbenches.
  • Data & NDA discipline: Mutual NDA available before you share a single spreadsheet or RFQ. Customer records stay inside your cloud tenant (AWS, Cloudflare, Azure, or GCP) and are never used to train third-party models.
  • Right-sized architecture: Roughly half of our recommendations use deterministic SQL, APIs, and clean UI screens with zero language-model calls when an LLM would add cost or error risk.

Direct contact

Send two sample RFQs, a redacted spreadsheet, or a note straight to [email protected], or start with the Free Workflow Snapshot for a written 1-page teardown within 2 business days.