Role summary
OryxenLabs is hiring an AI & Automation Engineer to build LLM integrations, RAG pipelines and n8n automations for our own products and for client projects. You will take features from a brief to production, and report to the founder at first, and to the Tech Lead once that role is filled.
OryxenLabs is a Pakistan-based venture studio building AI-native SaaS products and automation services. We work across four pillars: AI & Automation, Web Development, Branding and Consulting. You will build the AI and automation parts of our products (such as BidLenz, NexHR and our WhatsApp commerce agent) and of custom automation projects for SMEs in Pakistan and the GCC.
| Reports to | Founder & CEO for now; the Tech Lead once hired |
|---|---|
| Works with | Forward Deployed Engineer, BD / Sales Executive |
| Focus | LLM features, RAG, agents and workflow automation |
| Location | Pakistan — hybrid, with occasional GCC travel depending on the project |
| Type | Full-time |
Key responsibilities
LLM features
- Build features on Claude and other models through the API, with structured output that our code can parse and validate.
- Keep facts such as prices, stock and customer data in code and the database, not in the model.
- Add guardrails for agents that take actions, such as confirmation steps and human handoff.
Retrieval and data
- Build RAG pipelines on PostgreSQL and pgvector: chunking, embedding, retrieval and grounding answers in the client's own data.
- Turn PDFs, scans and spreadsheets into structured data that products can use.
Automation
- Build and maintain n8n workflows that connect products to WhatsApp, email, CRMs and business systems.
- Handle retries, rate limits and failures so workflows keep running without manual fixes.
Quality and delivery
- Test prompt and model changes against a set of real examples before they ship.
- Agree scope and estimates with the founder, and release to our Docker-based servers.
- Write short, clear notes on how each workflow and integration works.
Requirements
Must have
- 2+ years building production software, with code shipped in Python or TypeScript.
- Hands-on experience calling LLM APIs and getting reliable structured output.
- A working understanding of RAG: embeddings, vector search and why retrieval fails.
- Experience integrating REST APIs and webhooks, including error handling and retries.
- Solid SQL, ideally PostgreSQL.
- Clear written English for specs, notes and pull requests.
Nice to have
- Built workflows in n8n or a similar automation tool.
- Experience with the WhatsApp Business API or other messaging channels.
- Used pgvector, LiteLLM or an evaluation framework for prompts.
- Document parsing or OCR in a real project.
- Daily use of AI tools for building and coding, such as Claude Code, GitHub Copilot, OpenAI, OpenRouter, Hugging Face, TypeSafe AI's Jev and locally run models.
What success looks like
Targets will be agreed in your first month.
| Timeframe | Outcomes |
|---|---|
| First 30 days | You know our products, stack and deployment process. Your first change to an existing AI feature or n8n workflow is in production. |
| First 60 days | You own one AI feature or automation end to end, with a test set and written notes. Failures are caught and retried rather than fixed by hand. |
| First 90 days | You are shipping AI features for products and client projects on a regular rhythm. Prompt and model changes are checked against test examples before release. |
What we offer
- A competitive salary.
- Variable performance pay (VPP): an annual payout based on how the company and you performed over the year. It is paid in years when the company clearly exceeds its targets, so it is not guaranteed.
- A bonus when you exceed your own targets.
- Build AI-native products and automation used by real SMEs in Pakistan and the GCC.
- An early seat on the engineering team, with a say in how we build.
- Hybrid working.
- Paid access to AI and developer tools.
How to apply
Apply at oryxenlabs.com/careers. The online assessment takes about 90 minutes and includes a short AI engineering knowledge check, questions on the techniques behind our apps, two coding tasks in JavaScript or Python (one parses LLM output), a written design for a WhatsApp order bot, workplace scenarios and a short prioritisation game. Shortlisted candidates meet the founder face to face.