AI/ML Researcher · London
Raja Abraheem Rashid
AI/ML Researcher, working on nested learning, federated learning, and trustworthy AI. Building scalable, privacy-preserving intelligent systems across research and industry.
How I Work
My research and engineering work sits across three intersecting areas, foundational machine learning, large language models and agentic systems, and the infrastructure that lets ideas move from notebooks to production.
Machine Learning & Deep Learning
⚡ Designing deep learning architectures for medical imaging, time-series, and remote-sensing tasks
⚡ Building nested and multi-frequency learning frameworks for scalable, hierarchical representation learning
⚡ Developing physics-informed reinforcement learning systems with hard safety guarantees
⚡ Research spanning computer vision, NLP, reinforcement learning, and statistical learning theory
LLMs & Agentic AI
⚡ Fine-tuning LLMs using LoRA and QLoRA for domain-specific tasks under compute constraints
⚡ Building agentic multi-agent pipelines for autonomous financial analysis and digital forensics
⚡ Designing production-grade Retrieval-Augmented Generation (RAG) systems with hierarchical orchestration
⚡ Deploying quantised LLMs (INT8/INT4) for low-latency inference at national scale
MLOps, Distributed & Privacy-Preserving ML
⚡ Building distributed training pipelines using DeepSpeed and FSDP for large-scale model training
⚡ Designing federated learning and unlearning frameworks for privacy-preserving distributed intelligence
⚡ Containerising and orchestrating ML workloads with Docker and Kubernetes for reproducible research
⚡ End-to-end MLOps: CI/CD, experiment tracking with MLflow and Weights & Biases, data streaming with Kafka
Digitalisation with AI
Alongside my research, I founded Neura-X, a venture focused on building practical AI frameworks at the intersection of applied AI and enterprise digitalisation. Drawing on hands-on experience from Curium (autonomous-vehicle perception) and INK AI, Neura-X develops scalable solutions using frontier AI technologies such as retrieval-augmented generation, agentic pipelines, efficient transformers, and trustworthy ML systems. Our work is deliberately research-grounded: every framework and deployment is shaped by the same rigour that drives my academic publications. Alongside developing proprietary frameworks, we are also working with clients worldwide, from early-stage founders to established enterprises, helping them operationalise AI effectively without paying the hidden costs of poorly engineered machine learning systems.
Founder & Technical Lead
Agentic & LLM Systems
Production-grade RAG, multi-agent orchestration, fine-tuning (LoRA/QLoRA), and quantised inference.
Computer Vision & Perception
Industrial monitoring, multi-sensor calibration, and real-time CV pipelines built for edge deployment.
Trustworthy ML
Federated learning, privacy-preserving pipelines, and explainability evaluation under real-world drift.