AI Engineer

Summary

Join Pixalate, an online trust and safety platform, as an AI Research Engineer to bridge the gap between fundamental AI research and production systems. Lead cutting-edge research in agentic AI systems, multimodal analysis, and advanced reasoning architectures that directly impact millions of users. Work with the Research team to uncover fraud and national security threats, deploying innovations at scale. Lead research in autonomous agent systems, test-time compute optimization, and multimodal understanding applied to real-world challenges. You will design and implement multi-agent architectures, optimize inference-time compute, build advanced multimodal models, and research novel approaches to agent safety and alignment. The role offers the autonomy to pursue groundbreaking research and see your innovations deployed quickly.

Requirements

• PhD in Computer Science, AI, Machine Learning, or related field (or exceptional research track record)
• Published research in peer-reviewed venues demonstrating expertise in: Large Language Models and transformer architectures
• Agentic AI, autonomous systems, or multi-agent coordination
• Multimodal learning or computer vision
• Distributed systems and scalable ML
• Expert proficiency in Python and deep learning frameworks (PyTorch preferred, TensorFlow)
• Advanced experience with: Modern AI frameworks: LangChain, Hugging Face Transformers, Ray
• Agent development and orchestration
• RAG systems and vector databases
• Distributed training frameworks and GPU optimization
• Strong understanding of: Transformer architectures and attention mechanisms
• Reinforcement learning and reward modeling
• Neural architecture search and AutoML
• MLOps and production ML systems
• Track record of novel algorithm development and innovation
• Experience with large-scale experimentation and ablation studies
• Proficiency in research tools: Weights & Biases, MLflow, TensorBoard
• Strong theoretical foundation in optimization, statistics, and linear algebra

Responsibilities

• Design and implement multi-agent architectures for autonomous fraud detection and analysis
• Develop sophisticated agent coordination systems using frameworks like LangChain, AutoGen, or custom architectures
• Create tool-integrated AI agents capable of complex reasoning and decision-making
• Research novel approaches to agent safety and alignment in production environments
• Implement state-of-the-art reasoning systems inspired by recent breakthroughs (o1, DeepSeek-R1)
• Optimize inference-time compute allocation for complex analytical tasks
• Develop chain-of-thought and verification mechanisms for high-stakes decision making
• Research novel approaches to scaling reasoning capabilities efficiently
• Build advanced multimodal models for analyzing video, image, text, and behavioral data
• Develop sophisticated RAG (Retrieval-Augmented Generation) architectures: Design high-performance vector databases and hybrid search systems
• Implement advanced chunking strategies and semantic understanding
• Create context-aware retrieval mechanisms for complex documents
• Research cross-modal learning for fraud pattern detection

Preferred Qualifications

• Experience with fraud detection, cybersecurity, or trust & safety applications
• Contributions to open-source AI projects
• Industry research experience at leading AI labs (DeepMind, OpenAI, FAIR, etc.)
• Experience translating research into production systems
• Experience with: Mixture of Experts (MoE) architectures
• Constitutional AI and alignment techniques
• Efficient inference optimization (quantization, distillation)
• Real-time streaming ML systems

Benefits

• 25 days holiday plus Bank holidays
• Defined contribution Pension scheme
• Monthly internet reimbursement
• Casual, remote work environment
• Hybrid, flexible hours
• Opportunity for advancement
• Fun annual team events
• Being part of a high performing team that wants to win and have fun doing it
• Extremely competitive compensation