Production AI Deployment Engineer (LLM & Inference)

Pakistan Digital Authority; Islamabad; Presencial; PKR 1.800.000 - 3.600.000; Tempo integral

Resumo da vaga

Pakistan Digital Authority is seeking an AI Model Deployment Engineer to select, adapt, evaluate and deploy AI models for government use cases. The role translates user requirements into reliable production deployments, balancing model quality, response time, infrastructure capacity and cost.

You will work with product teams to define requirements, prepare data, fine-tune models, and deploy on inference servers with proper access controls.

Qualificações

• Bachelor’s degree in Computer Science, Artificial Intelligence, Software Engineering, Data Science, or a related discipline.
• At least 4 years of relevant AI/ML engineering experience, including at least 1 year of hands-on model deployment within that period.
• Demonstrated experience taking at least one AI/ML model through evaluation and deployment into a production environment.
• Proficiency in Python and practical experience with machine learning frameworks, Linux, containers and inference APIs.
• Hands-on experience with LLM evaluation, fine-tuning and GPU-based inference; ability to assess quality, latency and cost trade-offs.

Responsabilidades

• Work with product teams and users to define model requirements, acceptance criteria and performance targets for each use case.
• Compare candidate models using public benchmarks and representative use-case data; document model selection decisions.
• Prepare and clean fine-tuning data, verify data quality and authorised use, and maintain dataset provenance.
• Fine-tune models where needed using approaches such as LoRA and QLoRA, with reproducible configurations and versioned artifacts.
• Build evaluation datasets and assess accuracy, task performance and safety before and after model adaptation.
• Deploy models on inference servers such as vLLM, SGLang or Triton, with appropriate access controls and deployment documentation.
• Configure GPU memory, batching and context length; measure latency, throughput and cost per request.
• Apply quantisation methods such as AWQ, GPTQ or FP8 where supported, validating quality and performance before release.
• Work with MLOps and application engineers on release testing, monitoring, troubleshooting and rollback readiness.

Conhecimentos

Python

ML frameworks

Linux

Containers

Inference APIs

LLM evaluation

Fine-tuning

GPU inference

Troubleshooting

Documentation

Collaboration

Formação

Bachelor’s degree in CS/AI/SE/Data Science

Ferramentas

Kubernetes

vLLM

SGLang

Triton

Descrição da vaga

Pakistan Digital Authority is seeking an AI Model Deployment Engineer to select, adapt, evaluate and deploy AI models for government use cases. The role translates user requirements into reliable production deployments, balancing model quality, response time, infrastructure capacity and cost.

You will work with product teams to define requirements, prepare data, fine-tune models, and deploy on inference servers with proper access controls.