Senior Software Engineer

Location: Remote in Nepal, plus one office day a month with our Kathmandu team Hours: Romanian working hours (EET/EEST): about 12:45–21:45 Nepal time in winter, 11:45–20:45 in summer Engagement: full-time contractor On-call: none About us CloudHero is an AWS partner. Part of our team is already based in Nepal, and now we're building software products as well as running cloud work.

You'll be our first dedicated software engineer, working on: Our internal CRM. Pipeline, email sync with Gmail, AI agents that review deals, search, and an MCP server so AI assistants can work with the data. It's in production, and our sales team uses it every day. HeyLeave.com . An HR assistant on WhatsApp for companies in Romania.

Customer projects. Software we build for customers when they need it, scoped and delivered by you. How we build We use AI coding agents (Claude Code, Codex) every day, and they write much of the code. The skills that matter are the ones agents don't replace: Design. Choosing the simplest system that will hold up. Judgment.

Knowing when an agent is confidently wrong. Proof. Showing with tests and numbers that the software is correct and fast. So we don't hire for a language. We hire for fundamentals, architecture, and taste. You'll use whatever a product already uses, and you'll get up to speed quickly. Plan before code. Non-trivial work starts as a short written plan.

A reviewer (a person or another model) tries to break it, and the depth of review matches the risk: heavy for billing or data integrity, light for a copy change. Knowing when to stop reviewing is part of the job. Agents write drafts, and you own every line. You check their assumptions, challenge their suggestions, and decide what ships.

Tests prove claims. Concurrency, retries, tenant isolation and data integrity get tests that would fail if the claim were false. Evidence beats opinion. Measure, check the live system, then decide. Done means deployed, monitored, and documented , not just merged.

What you'll do

Act as product owner as well as engineer. Talk to users (our sales team, HeyLeave customers, project clients), turn vague requests into clear scope and acceptance criteria, prioritize the backlog, and say no to work that isn't worth it. Own the architecture. Choose data models and boundaries, decide how failures behave, and write the important decisions down.

Build end-to-end: APIs, frontend, background jobs, integrations, database migrations, deploys. Make it fast. Profile it, find the slow query, and prove the improvement with before-and-after numbers. Improve how we work. Add CI, speed up tests, and refine how we review AI-written code. You're the first engineer, so your habits become our standard.

Work with our DevOps team on infrastructure and monitoring. Your first 60 days Ship real changes to the CRM in your first two weeks, and get comfortable working in a large codebase you didn't write. Take product ownership of HeyLeave.com: understand its users, write a prioritized backlog with us, and ship the first improvements.

Find one measured performance or reliability problem in either product, fix it, and show the numbers. Add one improvement to how we build, such as CI, faster tests or a better review step. You are Senior. Several years building and running software in production. You've owned systems end to end and lived with your own decisions.

Product-minded. You care about what users need, not just what the ticket says. You can write a crisp spec and cut scope without losing the point. An architect. You choose the simplest design that works and can explain where it will break: under load, on a retry, on a crash halfway through, or with two users at once. In love with performance.

You've made something measurably faster and can tell the story with numbers. Serious about quality. You won't compromise on security or data integrity, and you make deliberate trade-offs everywhere else. Fluent with AI agents and LLM products. You use agents daily, know where they fail, and have ways to catch them. When an LLM sits in front of payroll data, you care that the numbers come from the source system and never from the model.

Passionate about building. Side projects are welcome but not required. We want to see work you're proud of, professional work included. A clear writer in English. Plans, specs, commit messages and release notes are most of the leverage when working with agents. Strong signals (not requirements) Deep database knowledge: transactions, locking, indexing, migrations on live data.

SaaS or multi-tenant products. Messaging bots (WhatsApp Business API), HR or payroll integrations, GDPR, email integrations, AWS. You've built something from zero to paying users. Hiring process (about 4 hours over 2–3 weeks) Intro call with HR (30 min). The role, the hours, contract terms, and why you're interested. Technical deep-dive (60 min) with the hiring manager and one of our engineers.

Walk us through something you built: why it's designed that way, what broke, and what you'd change. Then we'll review a short piece of code together. Live build session (90–120 min, online). You'll scope a small feature with us playing the user, plan it, then build it with AI agents while we watch, followed by a short debrief.

We provide the environment and agent access. We care about how you think, plan, and check your work, not how many features you finish. Present the offer (30 min, online).