Mid-Level Python AI Developer

Mid-Level Python AI Developer

Experience: 5–9 years

Role Overview

Independently implement Python AI components and LLM applications, from data preparation and model adaptation through evaluation, inference and production integration. Understand how models work as well as how to use them.

Key Responsibilities

• Build maintainable Python code, data pipelines, backend services and APIs for AI applications.
• Prepare and validate datasets, create appropriate train/validation/test splits, prevent leakage and track data and experiment versions.
• Implement training and fine-tuning workflows for suitable existing models, including supervised or parameter-efficient fine-tuning with guidance on broader architecture decisions.
• Run reproducible experiments, inspect loss curves and metrics, diagnose errors and improve model quality.
• Build LLM inference and application pipelines using tokenization, embeddings, retrieval, vector search and RAG where appropriate.
• Develop evaluation datasets and automated checks for task performance, groundedness and hallucinations; compare changes against a baseline.
• Integrate models through APIs and contribute to deployment, monitoring, latency and cost optimization.
• Write tests, use version control, document decisions and protect confidential data in datasets, prompts, logs and outputs.

Required Skills

• Strong Python development and data processing skills, with practical ML/deep learning implementation experience.
• Understanding of optimization, loss functions, overfitting and validation, plus tokenization, embeddings, attention and transformer architecture fundamentals.
• Ability to distinguish pretraining from training/fine-tuning and inference, and explain when model adaptation or retrieval is appropriate.
• Hands-on use of PyTorch or a comparable framework, transformer tooling, model evaluation and LLM application development.
• Experience with APIs, automated testing and reproducible software and experiment workflows.

Preferred Skills

• Experience with parameter-efficient fine-tuning, vector databases, containerized deployment and MLOps/LLMOps tools.
• Familiarity with monitoring model quality and application performance in production.