AI/ML Engineer

Amex Hub; Islamabad; On-site; PKR 1,800,000 - 3,600,000; Full time

Job summary

Amex Hub in Islamabad is seeking an AI/ML Engineer to drive end-to-end ML/AI workstreams—from data prep and experimentation to production deployment and optimization. You will build and fine-tune NLP/LLM-based or computer vision models, design experiments, and develop RAG pipelines with embeddings and vector databases.

Collaboration with the AI/ML Lead and other engineers is essential as you stay current with new models and tools.

Qualifications

• 2–5 years hands-on experience building and training ML/AI models.
• Strong Python skills with PyTorch and/or TensorFlow, plus Scikit‑learn, NumPy, and Pandas.
• Solid grasp of ML fundamentals (supervised/unsupervised, overfitting, regularization, metrics) and deep learning concepts.
• Practical experience with LLMs/NLP and/or computer vision.
• Hands-on experience in at least two areas: fine-tuning pretrained models (LoRA/PEFT), building RAG pipelines with embeddings/vector databases, deploying models via APIs.

Responsibilities

• Own ML/AI workstreams end-to-end: data prep, experimentation, model training, evaluation and deployment.
• Build, train, and fine-tune models for NLP/LLM-based or computer vision use cases based on project needs.
• Design and run experiments, track metrics, and iterate based on results.
• Build and maintain RAG pipelines with embeddings and vector databases; assess suitable approaches (fine-tuning vs prompting vs RAG).
• Implement evaluation frameworks to validate model quality before/after deployment.
• Deploy models to production via APIs and cloud infrastructure with attention to latency and reliability.
• Debug data, training stability, performance issues across the ML pipeline; write clean, tested code and participate in reviews.
• Contribute to architecture discussions with AI/ML Lead and engineers while driving your own workstreams.
• Stay current with new models, techniques, and tools and bring relevant ones to the team.

Skills

Python

PyTorch

TensorFlow

ML fundamentals

NLP

Computer vision

RAG pipelines

API development

SQL

Linux

Tools

Hugging Face Transformers

LangChain

Pinecone

Weaviate

Job description

Build and ship AI/ML solutions, from experimentation and model development through production deployment and optimization.

Engineering Bahria Town Islamabad Full-time

About the role

Responsibilities

• Own individual ML/AI workstreams end-to-end: dataset preparation, experimentation, model training,
• evaluation, and deployment
• Build, train, and fine-tune models for NLP, LLM-based, or computer vision use cases, depending on
• project needs
• Design and run experiments: select approaches, track metrics, and iterate based on evaluation results
• Build and maintain RAG pipelines, including embeddings, retrieval, and vector databases
• Assess whether fine-tuning, prompting, or RAG best fits a given problem, and recommend an approach
• to the AI/ML Lead
• Implement evaluation frameworks and benchmarks to validate model quality before and after
• deployment
• Deploy models to production through APIs and cloud infrastructure, with attention to latency and
• reliability
• Debug issues across the ML pipeline: data quality, training instability, model performance, and
• Write clean, tested, and documented code, and take part in code reviews
• Contribute to architecture discussions with the AI/ML Lead and other engineers while driving your own
• workstreams
• Stay current with new models, techniques, and tools, and bring relevant ones back to the team

Required Skills & Experience

• 2–5 years of hands‑on experience building and training ML/AI models
• Strong Python skills with PyTorch and/or TensorFlow, plus Scikit‑learn, NumPy, and Pandas
• Solid grasp of ML fundamentals (supervised and unsupervised learning, overfitting, regularization, evaluation metrics) and deep learning concepts
• Practical experience with LLMs/NLP and/or computer vision
• Hands‑on experience in at least two of the following, with working familiarity in the rest: (a) fine‑tuning

pretrained models, including parameter‑efficient methods such as LoRA/PEFT; (b) building RAG pipelines with embeddings, semantic search, and vector databases; (c) deploying models to production through APIs

• Experience with LLMs (open‑source and API‑based), prompt engineering, and model selection
• Working knowledge of SQL, Linux, and Bash/shell scripting
• Comfortable with Git, API development, and core cloud concepts (storage, compute, deployment) on AWS, GCP, or Azure
• Demonstrated ability to take a problem through dataset, experiment, model, evaluation, and deployment

Preferred Skills & Tools

• MLOps practices: experiment tracking, CI/CD for ML, and model monitoring (MLflow, Weights & Biases, Kubeflow, or similar)
• Training or inference on cloud GPUs (CUDA)
• LLM tooling: Hugging Face Transformers; OpenAI, Anthropic, or Gemini APIs; LangChain, LlamaIndex, or LangGraph
• Vector databases: Pinecone, Weaviate, FAISS, pgvector, or Milvus
• Agentic or multi‑step LLM workflows, including tool calling
• Computer vision with OpenCV and modern vision models
• Personal or open‑source projects that show independent, end‑to‑end ML work

What Makes Someone a Strong Fit

We value engineers who have trained, evaluated, and debugged real models, not only integrated existingAI APIs into an app. The ideal candidate can take a loosely defined problem and work it through to adeployed solution, asking for guidance on judgment calls rather than at every step. Engineers at the earlier end of the experience range will get more support from the Lead as they grow into full workstreamownership. Curiosity and a drive to keep learning matter as much as current skill level.

What we’re looking for

• Strong Python skills with PyTorch and/or TensorFlow
• Practical experience with LLMs/NLP and computer vision
• Experience with LLMs (open‑source and API‑based)
• MLOps practices: experiment tracking, CI/CD for ML, and model monitoring