ROLE SUMMARY

We are hiring an AI Engineer who builds and ships AI-powered systems in production — not someone who experiments in notebooks and calls it delivery. You will own the design and implementation of AI features across client projects: automation workflows, LLM integrations, fine-tuned models, RAG pipelines, and intelligent agents. You will also be expected to use AI tooling aggressively to accelerate your own development speed.

If your experience is limited to calling the OpenAI API and wrapping it in a UI, this is not the right role. We need someone who understands what happens inside the model, can build around its limitations, and can architect systems that are reliable, observable, and maintainable in production.

WHAT WE'RE LOOKING FOR

LLMs & Generative AI

• Deep, hands-on experience with LLM APIs — OpenAI, Anthropic, Gemini, Mistral, or equivalent — beyond basic completions: function calling, structured outputs, context management, and cost optimisation
• Experience designing and building Retrieval-Augmented Generation (RAG) pipelines: chunking strategies, embedding models, retrieval tuning, and re-ranking
• Practical experience fine-tuning or instruction-tuning open-source models (LLaMA, Mistral, Falcon, or similar) using LoRA, QLoRA, or full fine-tuning workflows
• Understanding of prompt engineering at a technical level — not just writing prompts but building prompt pipelines, evaluation harnesses, and systematic iteration workflows
• Familiarity with AI agent frameworks: LangChain, LlamaIndex, CrewAI, AutoGen, or equivalent — and knowing when NOT to use them

Automation & Workflow Engineering

• Hands-on production experience with n8n — building, deploying, and maintaining complex multi-step automation workflows; self-hosted setup and maintenance is a plus
• Hands-on production experience with Make.com (formerly Integromat) — scenario design, error handling, data transformation, and API integrations
• Ability to design automation architectures that are robust, observable, and recoverable — not brittle chains that break silently
• Experience connecting AI models into automation pipelines: LLM-driven routing, dynamic content generation, and intelligent decision nodes within workflows
• Familiarity with webhook design, async event handling, and queue-based architectures for automation at scale

Model Training & ML Engineering

• Practical experience training or fine-tuning models — not just using pre-trained ones. You should understand data preparation, training loops, evaluation metrics, and iteration cycles
• Solid understanding of core ML concepts: supervised and unsupervised learning, embeddings, attention mechanisms, tokenisation, and model evaluation
• Experience with training frameworks: PyTorch, Hugging Face Transformers, or TensorFlow
• Ability to evaluate model quality rigorously — building eval datasets, defining task-specific metrics, and detecting regressions
• Awareness of model deployment considerations: latency, memory footprint, quantisation, and serving infrastructure

Vector Databases & Semantic Search

• Production experience with at least one vector database: Pinecone, Weaviate, Qdrant, pgvector, or Chroma
• Understanding of embedding models (OpenAI, Cohere, sentence-transformers) and how embedding quality directly impacts retrieval performance
• Experience designing semantic search and similarity systems — including hybrid search combining vector and keyword retrieval
• Familiarity with indexing strategies, namespace management, and metadata filtering at scale

Python & Software Engineering

• Python in depth — not just scripting. Clean, modular, well-structured Python code with proper dependency management, testing, and packaging
• Experience building production Python services: FastAPI or Flask APIs, async programming with asyncio, and proper error handling and logging
• Strong understanding of data manipulation libraries: NumPy, Pandas, and familiarity with data pipeline tooling
• Version control discipline with Git — branching, code reviews, and maintaining shared AI/ML codebases with other engineers
• Ability to write readable, reproducible code that another engineer can maintain six months later

AWS & Infrastructure

• AWS experience relevant to AI workloads: EC2 (GPU instances), S3 for model artefact and dataset storage, Lambda for serverless inference, SageMaker for managed training and deployment
• Ability to deploy and serve AI models in production: containerised inference with Docker, model versioning, and endpoint monitoring
• Familiarity with cost management for AI infrastructure — GPU compute and API token costs can spiral; you should have a track record of keeping them under control
• Experience with CloudWatch or equivalent for logging, alerting, and observability of AI services in production

AI-Accelerated Development

• Active, practised use of AI tools in your daily development workflow — Cursor AI, GitHub Copilot, Claude, ChatGPT, or equivalent. This must be a real habit, not a talking point
• Ability to use AI tooling for code generation, debugging, documentation, and architecture review — and critically evaluate the output before it ships
• Demonstrated track record of shipping faster because of AI tooling, not in spite of it

Problem Solving

• Ability to diagnose and resolve failures in non-deterministic systems — AI outputs are probabilistic; you need structured approaches to debugging, evaluation, and improvement
• Strong analytical instincts when designing AI systems: trade-offs between accuracy and latency, cost and quality, simplicity and capability
• Comfortable operating in ambiguity — AI engineering involves undefined problems, unclear success criteria, and iterative discovery. You should thrive in that environment, not stall
• Track record of taking AI proof-of-concepts to production — with all the hardening, monitoring, and failure-mode handling that entails

WHAT YOU WILL OWN

• End-to-end design and delivery of AI features across client projects — from architecture to deployed, monitored production system
• Building and maintaining automation workflows in n8n and Make.com, including LLM-powered nodes and multi-system integrations
• Developing and maintaining RAG pipelines, fine-tuned models, and semantic search systems for client applications
• Integrating AI capabilities into existing product stacks — working closely with full stack engineers to expose AI functionality through clean APIs
• Evaluating and selecting appropriate AI tools, models, and infrastructure for each use case — with a clear rationale, not cargo-culting
• Setting up observability and evaluation frameworks for AI systems in production — monitoring quality, latency, cost, and failure rates
• Staying current with the AI landscape and bringing relevant advances to the team with a concrete proposal for how to apply them

GOOD TO HAVE

• Experience with multimodal models — vision, audio, or document understanding
• Familiarity with MLOps tooling: MLflow, Weights & Biases, DVC, or similar for experiment tracking and model management
• Exposure to graph-based AI architectures or knowledge graphs
• Experience with voice AI: STT/TTS pipelines, real-time audio processing
• Prior experience working directly with international clients on AI product requirements
• Open-source contributions or a public portfolio of shipped AI work

WHAT WE EXPECT FROM YOU

You will be onsite in Lahore. AI engineering is still a field where most practitioners overstate their depth — we will test yours thoroughly in the hiring process. We expect you to know the difference between a demo and a production system, and to hold yourself to the latter standard on every project you touch.

The field moves fast. We expect you to move with it — not just follow release notes, but evaluate new tools critically, experiment quickly, and apply what is genuinely useful to our client work.