Research Scientist

Research Scientist

Join a world-class research team advancing the frontiers of logical reasoning, transparency, and consistency in AI. This role combines deep scientific exploration with hands-on system building, offering the opportunity to design, prototype, and deploy neurosymbolic architectures that bridge neural and symbolic intelligence.

What You'll Do:

• Stay up-to-date with recent AI research and ideate novel experiments in hybrid neurosymbolic architectures progressing our efforts on logical reasoning, consistency and transparent AI systems.

• Lead end-to-end research projects: from ideation and literature review to working prototypes.

• Work alongside Research Engineers and Software Engineers to deploy large scale systems in production.

• Benchmark our models on public datasets, generate synthetic datasets, and set up data labelling experiments.

• Write research papers for major AI conferences, technical blog posts and assist with patenting our tech.

• Write high-quality, robust code that integrates neural and symbolic components.

• Collaborate with a team of scientists and engineers, articulating complex ideas clearly to technical and non-technical audiences.

What We're Looking For:

• Deep learning expertise with 4+ years of academic/industrial experience with Large Language Models, Fine-tuning open LLMs, Reinforcement Learning, RAG, agentic systems and neurosymbolic representations.

• Experience with Python and common AI Frameworks like Hugging Face Transformers & TRL, PyTorch, DeepSpeed, DSPy, LangChain, LangGraph.

• Track record working as an independent contributor capable of end-to-end R&D with a deep understanding of LLM architectures, training paradigms, demonstrable experience in utilising and deploying LLMs based agentic systems.

• A significant publication track record in top AI conferences like NeurIPS, ICML, ICLR, AAAI, ACL, etc. or equivalent experience writing technical reports, blog posts, patents in the industry.

• Enthusiasm to learn and get up to speed with cutting-edge technologies which you may not already be deeply familiar with.

• Excellent verbal and written communication skills with a proven track record of mentoring and coaching others.

• Capable of working collaboratively and proactively in a fast-paced environment with scientists, engineers, and non-technical stakeholders.

Desirable:

• Use of Python libraries that encourage best practices such as uv, pytest, pylint, black etc.

• Experience with program synthesis and structured representations.

• Start-up experience.

• Git/Github.

• Proficiency working with cloud platforms for training and deploying hybrid AI systems.

What's in it for You

Joining UnlikelyAI means working on technology that pushes boundaries while shaping your career:

• Work on cutting-edge AI projects with both customer-facing and internal impact.

• Be part of a world-class, collaborative team with huge ambitions.

• Explore novel challenges in a fast-moving environment where creativity thrives.

• Help shape the foundational layers of a new paradigm in AI.

Working at UnlikelyAI

We offer a range of benefits and perks designed to support our team's wellbeing and work-life balance:

• We have a hybrid working arrangement, flexibly balancing working from home and office-based working

• Our office is located in Bloomsbury, approximately a five minute walk to both Tottenham Court Road and Holborn stations

• We provide free team lunches every Tuesday, Wednesday and Thursday

• We schedule a variety of optional social and extra-curricular activities

• We have an annual offsite, usually to an international location, where we can work and socialise in the sun

Equal Opportunities

We are committed to having a truly diverse team where everyone is encouraged to be their authentic selves. We, therefore, do not discriminate in employment based on gender, race, religion, sexual orientation, national origin, political affiliation, disability, age, marital status, medical history, parental status or genetic information. Having a broad mix of people helps us to be the best we can.