SoTalent · 1 day ago
Artificial Intelligence Researcher
SoTalent is seeking an AI Researcher who is passionate about advancing AI and applying cutting-edge research to solve real-world problems. The role involves conducting applied research to develop and optimize large-scale AI models, translating research insights into production-ready solutions that deliver measurable business impact.
Responsibilities
Conduct applied research to develop and optimize large-scale AI models, including foundation models and LLMs
Design, train, evaluate, and deploy models across language, vision, graphs, and sequential data
Explore state-of-the-art techniques in self-supervised learning, robustness, explainability, RLHF, and more
Work with advanced AI stacks such as PyTorch, HuggingFace, Lightning, AWS Ultraclusters, VectorDBs
Translate research insights into production-ready solutions that deliver measurable business impact
Qualification
Required
PhD (or Master's with research experience) in Computer Science, AI, Machine Learning, Mathematics, or related field
Strong programming skills in Python, Go, Scala, or Java
Deep understanding of AI fundamentals and experience with large-scale deep learning models
Proven ability to publish or contribute to impactful research in top-tier conferences (e.g., NeurIPS, ICML, ICLR, ACL)
Ability to define and execute a research agenda, from problem selection to implementation
Preferred
Large Language Models: Pretraining, finetuning, optimization, and scaling (10B+ parameters)
Graph & Sequential Models: GNNs, time-series, recommender systems, and large-scale graph modeling
Optimization: Model sparsification, quantization, parallelism, gradient checkpointing, and compiler-level improvements
Contributions to open-source frameworks (e.g., PyTorch Geometric, DGL)
Experience deploying research-driven models in production environments
Company
SoTalent
At SoTechTalent, we specialise in connecting forward-thinking tech companies with world-class talent.
Funding
Current Stage
Early StageCompany data provided by crunchbase