Junior AI Engineer – Build LLM‑Powered Applications

About the role

We are looking for a Junior AI Engineer to join our team and help design, develop, and deploy AI‑driven solutions. You will work on cutting‑edge language‑model applications, experiment with model optimization, and contribute to real‑world products.

Key responsibilities

• Develop AI‑powered applications using large language models (LLMs), Retrieval‑Augmented Generation (RAG) and Model Context Protocol (MCP).
• Implement function calling, agent orchestration and tool integration features.
• Benchmark models with LLM‑as‑a‑Judge, few‑shot evaluation and automated testing pipelines.
• Apply model optimization and quantization techniques.
• Containerize and deploy services with Docker and Docker Compose.
• Use Hugging Face Transformers for embeddings, tokenization and text classification.
• Work with vector databases for semantic search and retrieval.
• Apply computer‑vision methods (segmentation, feature extraction, object detection) using OpenCV.
• Fine‑tune deep‑learning and machine‑learning models.
• Integrate applications with relational (MariaDB, PostgreSQL) and non‑relational databases.
• Handle background processing with queuing and task‑orchestration frameworks.

Required profile

• 1–2 years of hands‑on AI/ML engineering experience.
• Strong proficiency in Python.
• Familiarity with PyTorch or TensorFlow.
• Good understanding of core ML/NLP concepts, especially LLMs and RAG workflows.
• Experience working in Unix/Linux environments.
• Exposure to asynchronous frameworks such as FastAPI or asyncio.
• Comfort with Docker and containerised development.
• Practical experience with Git version control.

Required skills

• Python
• PyTorch
• TensorFlow
• Docker & Docker Compose
• FastAPI
• asyncio
• Git
• Hugging Face Transformers
• OpenCV
• MariaDB
• PostgreSQL
• Vector databases (e.g., Qdrant, ChromaDB)
• RabbitMQ or Redis Queue
• LLMs, Retrieval‑Augmented Generation, Model Context Protocol
• Function calling, agent orchestration, tool integration
• LLM‑as‑a‑Judge, few‑shot evaluation, automated testing pipelines
• Model quantization and optimization
• Semantic search, image segmentation, object detection, fine‑tuning deep‑learning models