Founding AI Research Scientist

Blue Wolf Digital; Greater London; Hybrid; GBP 120,000 - 170,000; Full time

Job summary

Blue Wolf Digital is seeking a Founding AI Research Scientist in London with a hybrid work mode. You will lead the design, training, and refinement of cutting-edge LLMs and deep-learning models, building a scalable research stack alongside the founder.

The role emphasizes ownership, ambitious research goals, and the ability to drive the full model lifecycle from data to deployment in a fast-paced, founder-led environment.

Qualifications

• Experience fine-tuning or pretraining LLMs and deep-learning models.
• Fluency in PyTorch and end-to-end training experience (data → deployment).
• Strong research ambition and willingness to take ownership in ambiguous environments.
• Founder mindset: ownership, pace, discretion in demanding work.
• Eligibility for UK security clearance is advantageous.

Responsibilities

• Design, train and fine-tune custom deep-learning models and proprietary LLMs.
• Own a detection and analysis pipeline end-to-end, from raw signal to client-ready insight.
• Push matching and extraction beyond similarity search to precision you engineer, on models you train.
• Turn proprietary data into models that beat commercial baselines.
• Set research strategy directly with the founder, then hire and lead the research team.

Skills

LLMs

Deep learning

Fine-tuning

Pretraining

End-to-end training

Tools

PyTorch

Job description

Founding AI Research Scientist | LLMs · Deep Learning · Frontier AI

London (Hybrid)

My customer builds frontier AI that gives an information advantage to those solving the world's hardest problems - working with governments, national security, and private organisations in the most demanding environments.

This is a founder-led research role. The CEO (AI PhD) has already built the model stack powering what the company ships today. You're the hire who takes it further - scaling it into a team and owning the entire stack.

What you’ll do:

• Design, train and fine-tune custom deep-learning models and proprietary LLMs - architecture, training objectives, evaluation
• Own a hard detection and analysis pipeline end-to-end, from raw signal to client-ready insight
• Push matching and extraction beyond similarity search to precision you engineer, on models you train
• Turn proprietary data - at a scale most labs never see - into models that beat commercial baselines
• Set research strategy directly with the founder, then hire and lead the research team

What you’ll bring:

• Experience fine-tuning or pretraining LLMs and deep-learning models
• Fluency in PyTorch, comfortable owning training runs end-to-end (data → deployment)
• Research ambition above all - you reach for the approach that could win, not just the one that works
• A founder mindset: ownership, pace, comfort with ambiguity, and the discretion this work demands
• Eligibility for UK security clearance is advantageous

Why this role:

• A proprietary dataset at a scale you can't get anywhere else, aimed at a mission that matters
• Real ownership and founder-level access - the freedom to do genuine research