LLM & Generative AI Engineer

Genixbit; Greater London; On-site; GBP 70,000 - 100,000; Full time

Job summary

Genixbit in London is seeking an AI Engineer to fine-tune open-source models for domain tasks, improve deployment pipelines, and develop context management and semantic search solutions.

The role requires hands-on experience with PyTorch, Hugging Face Transformers, and parameter-efficient fine-tuning, plus deploying models with vLLM, Ollama, or Triton. Strong software engineering practices are essential.

Qualifications

• 3+ years of experience in NLP and Generative AI.
• Hands-on PyTorch, Hugging Face Transformers, and PEFT/LoRA experience.
• Experience deploying models with vLLM, Ollama, or Triton.

Responsibilities

• Fine-tune open-source models (Llama, Mistral, Qwen) for domain functions.
• Optimize deployment pipelines for low latency and high throughput.
• Build advanced context management and semantic search solutions.
• Implement prompt evaluation frameworks and guardrail architectures.

Skills

NLP / Generative AI

PyTorch

Hugging Face Transformers

PEFT / LoRA

Model deployment

Tools

vLLM

Ollama

Triton Inference Server

Job description

About the Role

Join our AI Engineering division in London to specialize in LLM fine-tuning, retrieval-augmented generation (RAG), and hosting private models. You will be responsible for tailoring deep learning models to specialized domain tasks.

Key Responsibilities

• Fine-tune open-source models (Llama, Mistral, Qwen) for specific domain functions
• Optimize model deployment pipelines for low latency and high throughput
• Build advanced context management and semantic search solutions
• Implement prompt evaluation frameworks and guardrail architectures

Requirements

• 3+ years of experience focusing on Natural Language Processing and Generative AI
• Hands-on experience with PyTorch, Hugging Face Transformers, and parameter-efficient fine-tuning (PEFT/LoRA)
• Experience deploying models with vLLM, Ollama, or Triton Inference Server
• Strong background in software engineering best practices and clean code