AI Model Deployment Engineer

Role Purpose
The Pakistan Digital Authority (PDA) 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.

Key Responsibilities

• 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.

Required Qualifications & Experience

• Bachelor’s degree in Computer Science, Artificial Intelligence, Software Engineering, Data Science, or a related discipline.
• At least 4 years of relevant experience in AI/ML engineering or model development, 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, model libraries, 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.
• Strong analytical, troubleshooting, documentation and collaboration skills.

Preferred Experience

• Experience with government or enterprise AI applications, multilingual models, or Urdu-language evaluation.
• Exposure to Kubernetes, inference scaling, model quantisation and secure deployment environments.
• Understanding of data privacy, model licensing, information security and responsible AI practices.