ML Engineer (AI/NLP, Vector Search) – 6-month contract
Our client, is a UK-based tech company that helps startups discover and secure funding opportunities – from venture capital to alternative sources – tailored to their specific needs. The funding landscape is notoriously opaque for most startups, which our client addresses by using AI to scrape and analyse relevant funding data, then recommending the most suitable options.
They are seeking a highly skilled and motivated Machine Learning Engineer to join their team. The ideal candidate is proficient in Python and has experience working with modern AI (NLP) tools and workflows.
This role requires expertise in building and deploying scalable vector search solutions, installing and working with large language models (LLMs) like Llama 3 in cloud solutions, and developing pipelines for summarization and embedding generation.
Apply if you enjoy:
- A fast-paced startup environment
- Building AI / ML products and staying ahead of AI developments
- Working in small, lean teams, and happy to roll your sleeves up to get the work done
- Co-ideating entire solutions to problems, rather than coding small tickets
What you will be doing:
You will be in a team of four engineers reporting to the CIO, owning the co-ideation and creation of end-to-end solutions. You will be one of two experienced Python engineers, and your focus will be most heavily on: Updating existing pipelines with LLM-based techniques the existing data pipelines. Using same/ related techniques to build new pipelines for new use cases.
These pipelines are used in the company recommendation engine for startups, and will be used with your improvements for processing structured startup data for end-users.
Our client has recently Installed and fine-tuned Llama 3 for use in summarisation and is in the process of replacing OpenAI in its pipelines. This first work with in-house LLM is nearing completion at the time of writing.
Data Engineering & Programming:
o Develop and maintain robust data ETL pipelines using Python.
o Leverage Docker for containerized development and deployment workflows.
LLM Integration:
o Install, fine-tune Llama 3 and other large language models in cloud solutions e.g. Sagemaker, Vertex. Note: using ChatGPT will not suffice.
o Develop custom workflows to generate outcomes using LLM e.g. text summarization, embeddings, classifications.
o Complete PEFT to fine-tune models.
o Conversant with tools for e.g. NER.
o Using latest enablers from e.g. HuggingFace.
Model Training & Fine-tuning:
o Train and adapt LLMs to specific business requirements using domain-specific datasets.
o Evaluate and improve model accuracy, scalability, and efficiency.
Vector Search Solutions:
o Design and implement vector-based search systems using Pinecone or similar technologies.
o Optimise search performance for large-scale datasets to support real-time and batch querying.
Documentation:
o Document workflows, codebases, and best practices to ensure maintainability and scalability.
Requirements
Problem-Solving:
o Strong analytical and troubleshooting skills.
o Proven ability to design scalable and efficient solutions for complex AI/ML problems.
AI/ML Expertise:
o Strong proficiency in Python with experience in libraries such as Pandas, NumPy, and PyTorch.
o Hands-on experience with Docker for development and deployment.
o Familiarity with CI/CD tools (e.g., Jenkins, GitHub Actions, GitLab CI).
o Familiarity with cloud platforms (AWS, GCP, or Azure) for deploying ML workflows
o Strong understanding of NLP techniques
o Experience working with vector databases, particularly: Pinecone.
o Understanding of LLM installation/configuration like Llama 3 in Vertex.
o Ability to fine-tune (PEFT, LORA) models for tasks such as summarisation, text generation, and embedding creation.
Benefits
- Competitive compensation structure.
- Build a sector-defining product to meaningfully drive funding into the global startup ecosystem.
- Flexible remote work setup.
- Room for professional growth and skill development.
- Collaborative and inclusive team culture that values everyone’s input.