Junior Software Engineer

Who we are

Pathos Communications is a PR consultancy whose real IP is software. We build Pressella — an AI outreach and conversation platform (SMS, WhatsApp, email, voice) that engages prospects at scale and books meetings for our PR Consultants — and Pathos Mind, our internal portal for lead management, automated research, and AI article generation. Production systems, hundreds of thousands of contacts, real revenue.

The role

This is AI-first engineering: Claude Code (or similar) does most of the typing; you work a level up, directly with the CTO. You will:

 - Turn loose product ideas into precise PRDs and specs — with AI doing the implementation, the spec is the engineering.

 - Direct AI agents to build Pressella and our internal tools — decompose the work, write the prompts, review and test what comes back, ship it.

 - Build the systems that build software — agent workflows, evals, guardrails, CI.

 - Own quality — read code you didn't write, catch the subtle bug, debug production, leave things safer than you found them.

Stack & how we work

Python/FastAPI and TypeScript/Next.js on GCP (Cloud Run, Firestore, BigQuery, Terraform), Anthropic and OpenAI APIs, Twilio and HubSpot integrations. Trunk-based development with AI-reviewed PRs: small TDD commits, branches live hours not days, main always releasable, incomplete features ship dark. You don't need every tool on day one — you need to get productive fast, with AI assistance.

You

 - Solid fundamentals in Python or TypeScript — you can read unfamiliar code and explain why it works.

 - Real experience driving development with AI coding tools (Claude Code, Cursor, or similar) — not just autocomplete.

 - A clear writer: specs, commit messages, incident notes — writing is half this job.

 - Comfortable with cloud (GCP preferred) and Git; TDD experience or the appetite to learn it.

 - In Dubai or willing to relocate (visa sponsored).

We weigh shipped work and professional skepticism toward AI output over years of experience.

How to apply — build something

1. Use AI to build a simple app for a PR company — your choice of idea; that's part of the test.

2. Push it to a public GitHub repo — commit history counts.

3. Record a ~5-minute video: the idea, how you built it with AI, a live demo, what went well, and what you'd do next.

Keep it small and finished — an evening or two is plenty. We're assessing spec thinking, how you direct AI, whether you verified the output, and how clearly you communicate.

Send repo + video via DM to https://www.linkedin.com/in/scottdfeltham/

Note: Failure to complete the task will exclude you from next stage.