Lead Machine Learning Engineer – Conversational AI Platform
Location: Remote / Flexible
Contract Type: Permanent
We’re partnering with a pioneering technology business that’s redefining how humans interact with AI at scale. Their platform powers millions of personalised conversations every day, and they’re looking for a hands-on technical leader to drive the next evolution of their large language model (LLM) capabilities.
This is a rare opportunity to take ownership of the architecture, optimisation, and deployment of cutting-edge models in a high-growth environment. You’ll lead a small, highly skilled team while remaining deeply involved in coding and problem-solving.
Key Responsibilities
* Lead from the front: Architect solutions, mentor engineers, and contribute production-level code (Python/PyTorch).
* Optimise the conversational engine: Enhance context handling, retrieval pipelines, and inference speed for real-time performance.
* Own the model lifecycle: Define strategies for fine-tuning, RLHF/DPO, and evaluation; oversee data sourcing and curation for improved steerability and diversity.
* Design advanced safety systems: Build classifiers and moderation frameworks that go beyond binary decisions, ensuring compliance without compromising user experience.
What we’re looking for
* Proven delivery at scale: 6+ years in engineering, with significant experience shipping ML/LLM solutions to large audiences.
* Technical depth: Strong Python/PyTorch skills and familiarity with modern LLM stacks (e.g., Hugging Face, vLLM, fine-tuning pipelines).
* Practical alignment expertise: Understanding of how prompts, sampling, and architecture influence model behaviour.
* Execution mindset: Ability to prioritise shippable solutions over academic perfection.
* Ownership mentality: Committed to performance, safety, and user experience beyond initial deployment.
The environment
* Team: Small, senior engineering group reporting into you, with close collaboration across DevOps and product.
* Culture: High autonomy, rapid decision-making, and direct access to leadership.
* Tech stack: Python, PyTorch, vLLM, Hugging Face, RAG pipelines, custom moderation systems.