Senior Principal AI/ML Engineer

Site Name: UK - London - New Oxford Street, UK - Hertfordshire - Stevenage
Posted Date: Dec

GSK is a global leader in pharmaceuticals and healthcare, with a relentless commitment to advancing healthcare for the betterment of humanity. Our mission is to help people around the world do more, feel better, and live longer. We achieve this by researching, developing, and providing innovative medicines and vaccines. Our dedication to scientific excellence and ethical practices guides everything we do.

R&D at GSK is highly data-driven, and we're applying AI/ML and data science to generate new insights, enable analytics, gain efficiencies and automation.

This role is based in an AI/ML team that is already working on projects involving Generative AI, Information Retrieval, NLP/NER/RE, document classification and data science, and has won awards and recognition for its work. The team's future projects will be in diverse areas, such as regulatory, clinical, legal and HR. Versatility is key, with an ability to quickly understand domain data and requirements and translate them into solutions. You will interact with architects, software and data engineers, modelers, data scientists, other AI/ML engineers, product owners as well as other team members in Clinical Solutions and R&D. You will actively participate in creating technical solutions, designs, implementations and participate in the relentless improvement of R&D Tech systems in alignment with agile and DevOps principles.

A total of three Senior AI/ML Engineer positions are available, and we're looking for demonstratable expertise across a selection of the following key competencies: Generative AI, model building, training and evaluation, natural language processing, classification problems and software development. You should also be versed in agile ways of working, source control and the Azure cloud.

In this role you will

You'll have the opportunity to work on a mixture of the following:

• Generative AI
  • Design and develop RAG based applications.
  • LLM fine-tuning, including preparation of training sets from internal data
  • Agent based applications
  • Evaluating use-case specific LLMs
• AI/ML
  • NLP: Named Entity Recognition across a variety of unstructured data.
  • Evaluating and training BERT-like models such as GLiNER, NuNER for NER tasks. Analysing trade-offs between these models and LLMs for NLP tasks.
  • Relationship Extraction: Evaluating different models for use-case specific RE, such as ATG.
  • Document and text Classification
• Data Science: Data clustering algorithms, feature engineering
• Evaluate and integrate new technologies and models.
• Cross-team collaboration, identifying innovations and architecting solutions.
• Provide leadership and technical direction to various business units and partners

Why you?

Qualifications & Skills:

We are looking for professionals with these required skills to achieve our goals:

• Bachelor's degree in computer science
• Extensive experience working in AI/ML
• Generative AI: Demonstratable experience of RAG, including chunking strategies, vectorising and indexing data, retrieval strategies and reranking, prompting strategies, function calling. Our current tech-stack is OpenAI, LangChain, Azure AI, Python, pg_vector, Sinequa. Experience of multi-agent systems (LangGraph, Autogen, CrewAI etc.) would be a plus, as would experience of multimodal LLMs (like GPT4 Omni, Qwen-vl, DocOwl etc.) for understanding complex documents and images.
• AI/ML: Hands on experience with training and evaluating BERT-like models in real-world applications, especially in NLP or classification problems.
• Hands on experience with ML tools like TensorFlow, PyTorch etc.
• Experience with data science libraries such as NLTK, Scikit-learn, SciPy, (Sci)SpaCy etc.
• Excellent problem-solving and programming skills in Python
• Excellent communication skills

Preferred Qualifications & Skills:

If you have the following characteristics, it would be a plus:

• Master's or PhD in Computer Science
• Some experience with MLOps would be very beneficial.
• Experience with building search applications using Azure Search, Sinequa, Elastic or anything Lucene-based would come in handy but is not necessary.
• Familiarity with Azure cloud (AKS, Azure AI, ADF, Document Intelligence etc.), though you'll work with experts to gain this skill as well.
• Experience in training, evaluating and hosting open source LLMs would be a major benefit.

Closing Date for Applications: Thursday 9th January 2024 (COB)

Please take a copy of the Job Description, as this will not be available post closure of the advert.
When applying for this role, please use the 'cover letter' of the online application or your CV to describe how you meet the competencies for this role, as outlined in the job requirements above. The information that you have provided in your cover letter and CV will be used to assess your application.