NLPatent is an industry leading AI-first patent research platform that was an early mover in the application of Large Language Models. We use a combination of proprietary and off-the-shelf machine learning models and NLP techniques to help our users answer patent related research questions such as "Is my invention patentable?". Increasingly, we're using generative LLMs to build agentic workflows that answer research questions without any human intervention.
As a staff level applied scientist at NLPatent, you'll be the technical lead for all data and NLP related work at NLPatent. You'll be responsible for improving search performance (precision and recall), optimizing search algorithms to ensure quick results, designing and implementing agentic workflows, building and maintaining data pipelines, and much more.
Requirements
- 8+ years of commercial engineering and data science experience
- Experience with technical leadership
- Extensive experience with Python programming
- Commercial Machine Learning experience
- Experience working in Natural Language Processing (NLP)
- Experience training machine learning models using Tensorflow and PyTorch
- Experience working with search databases (e.g. Elasticsearch, Opensearch, etc.)
- Experience with cloud service providers (e.g. AWS)
- Willing to work 3 days per week onsite in the office in London
Benefits
- Stock Option Plan
- 25 days holiday allowance
- Pension contributions
- Flexible Working
- Training & Development
Top Skills
What We Do
NLPatent is an industry leading AI-based patent search and analytics platform trusted by Fortune 500 companies, Am Law 100 firms, and research universities around the world. The platform takes an AI-first approach to patent search; it's built from a proprietary Large Language Model trained on patent data to truly understand the language of patents and innovation. Users simply describe their invention in full sentences and conceptually relevant results are generated instantly; consistently outperforming human experts on speed and accuracy. The system is simple, intuitive, and iterative. Best of all, it explains the relevant sections of each patent it identifies, removing the "black box" often experienced by other AI-based platforms.
