1 month ago
London, United KingdomSenior
Responsibilities
- Lead the design, quantization, fine-tuning, optimization, and deployment of SLMs and vision-language models on low-power edge runtimes.
- Design and maintain lightweight on-device graph databases and relationship-extraction pipelines for structured and unstructured sensor data.
- Implement bounding guardrails, prompt evaluation, anti-hallucination controls, and Dependable AI governance requirements under JSP 936.
- Build reproducible model-training, evaluation, and containerized deployment pipelines for air-gapped and low-bandwidth environments.
- Translate complex ML and AI concepts into technical recommendations for MoD stakeholders, DSTL assessors, and prime contractors.
Requirements
- Active UK SC Clearance and strong knowledge of AI safety, non-repudiation, human-in-the-loop constraints, and JSP 936 Dependable AI requirements.
- At least 3 years of production experience deploying ML models to edge runtimes.
- Experience with model quantization techniques including INT8, INT4, and AWQ, plus NPU/GPU execution acceleration.
- Proficiency in Python and the PyTorch/Hugging Face ecosystems.
- Knowledge of NLP, semantic summarization, graph-based data structures, graph databases, vector embeddings, and network analysis.
- Understanding of Protobuf, JSON, XML, and streaming analytics.
- Experience integrating ML runtimes into Android ART through Chaquopy, JNI, or native C++ libraries is desirable.
- Experience with military sensor feeds, signals intelligence, SIGRF, or Cursor-on-Target data is desirable.
- Publications or prior delivery with DSTL, DAIC, or Defence Innovation programs are desirable.
Benefits
- Hybrid working with a work-from-home equipment allowance.
- Annual salary review, pension starting at a 5% employer contribution and increasing up to 8%, and group life assurance.
- 25 days of annual leave plus bank holidays, with the option to buy or sell additional days.
- 10 paid days for Reservist Military Service and 2 paid volunteering days per year.
- Fully funded professional certifications, 5 days of paid study leave, a £500 annual Personal Choice learning fund, coaching, and team training.
- Vitality Private Medical Insurance, Apple Watch and gym rewards, and a Cycle to Work scheme.
- Candidates must have the right to work in the UK without sponsorship and have lived in the UK continuously for the last 5+ years.
- The role is based in London, United Kingdom, and requires active UK SC Clearance for government customer projects.
Categories
