Senior LLM Engineer - Multimodal Audio & Voice AI

Harnham; Slough; Hybrid; GBP 108,000 - 132,000; Full time

Benefits offered by this job

Hybrid work model

London-based office

Job summary

Harnham invites applications for a Senior AI Engineer to build production-grade LLM systems and shape a new AI Lab focused on next-generation AI-powered consumer experiences. You will work in a small, high-autonomy team inside a global organisation, collaborating with engineering, product and design on greenfield AI systems.

The role covers tuning tone and output quality, building retrieval and RAG workflows, developing audio-first products with TTS/STT, and shaping evaluation frameworks for

Qualifications

• Experience building production-grade LLM systems.
• Strong track record with retrieval and RAG workflows.
• Proficiency in Python and modern ML tooling.
• Hands-on experience with audio-first AI products and multimodal systems.

Responsibilities

• Build and deploy production-grade LLM systems.
• Design and optimise RAG and retrieval workflows.
• Fine-tune models for tone, style, and output quality.
• Develop audio-first AI products using TTS/STT workflows.
• Build multimodal AI experiences across text, voice, and search.
• Create evaluation frameworks for hallucination and response quality.
• Prototype and iterate rapidly in a greenfield environment.

Skills

LLM systems

Retrieval/RAG

Model fine-tuning

Audio-first AI

Multimodal AI

Evaluation frameworks

Python

HuggingFace

LangChain

TTS/STT

Tools

LangChain

RAG pipelines

TTS/STT tooling

Job description

Harnham invites applications for a Senior AI Engineer to build production-grade LLM systems and shape a new AI Lab focused on next-generation AI-powered consumer experiences. You will work in a small, high-autonomy team inside a global organisation, collaborating with engineering, product and design on greenfield AI systems.

The role covers tuning tone and output quality, building retrieval and RAG workflows, developing audio-first products with TTS/STT, and shaping evaluation frameworks for