Founding AI R&D Engineer
- Artificial Intelligence
- London
- Permanent
- £100000 - £200000
- #51022
-
About the role
Founding AI R&D Engineer (VLMs) – AI Games startup – London Hybrid upto £200k
Spent serious time fine tuning Vision Language Models, not just reading about them?
Worked with video data, temporal context, and models that need to understand what’s happening over time, not just in a single frame?
Comfortable taking research ideas and turning them into something that actually runs in production?
If yes, keep reading.
The challenge
This startup is building autonomous agents that understand and interact with games the way humans do. That starts with perception.
They’ve built their own proprietary model that has beaten the likes of Google and Anthropic on benchmarking. You’ll own the core VLM work that turns raw video and images into usable game understanding. This is not a pure research role and it’s not just application glue. It sits in the middle, applied R&D with real constraints.
Your models will be the brain that agents rely on in live environments.
What you’ll be working on
- Fine tuning and optimising Vision Language Models such as LLaVA, CLIP, Flamingo, Gemini or similar
- Training models on video data, not just static images
- Designing inputs that capture temporal context across multiple frames
- Balancing model quality with inference speed and cost
- Taking models all the way to deployment in a production environment
About you
You’re likely an AI Engineer or Applied Researcher who:
- Has hands on experience fine tuning Vision Language Models
- Has worked with video data and understands temporal modelling challenges
- Enjoys applied R&D and shipping code
- Can handle some MLOps to get models into production
- Thinks about inference, latency, and cost, not just accuracy
A research or academic background is helpful but not required. What matters is that you’ve actually built and deployed things.
Nice to have, not must haves
- CUDA or low level inference optimisation
- PEFT or model compression techniques
- Experience optimising models for real time inference
Not for you if
- You only want to publish papers
- You avoid production responsibility
- Your experience with VLMs is mostly theoretical
Interested?
If you’ve built agents that actually do things and want to take real ownership, get in touch.
- Click ‘Apply Now’
- mo@mbnsolutions.com
- 020 3773 6498
You must be eligible to work in the UK
