AI Researcher

About NeoSpace

NeoSpace is an innovative startup shaping the future of technology with cutting-edge artificial intelligence solutions. We build specialized AI models to streamline processes and transform our clients' experience. Our goal is to make people's lives simpler and companies more efficient by creating smarter, more accessible products and services.

What we're looking for

The AI Researcher (Deep Learning Modeling) will be responsible for investigating, developing, and validating new architectures, training techniques, and advanced methodologies for deep learning models. The role involves both applied and exploratory research, contributing to the scientific advancement of the field as well as to the development of innovative solutions aligned with the organization's product and strategy.

Responsibilities

• Conduct advanced research on modern neural architectures (Transformers, Diffusion Models, advanced RNNs, Graph Neural Networks, multimodal models, etc.).
• Develop and evaluate large-scale training algorithms and pipelines, including optimization, regularization, parallelization, and acceleration techniques.
• Formulate hypotheses, design experiments, and analyze quantitative and qualitative results with scientific rigor.
• Propose improvements to existing models, exploring new deep learning techniques, architectures, and paradigms.
• Build experimental prototypes and demonstrate their feasibility to technical stakeholders.
• Collaborate with engineering, product, and science teams to turn research into high-impact solutions.
• Document experiments, methodologies, and findings in technical reports and internal/external publications.
• Stay on top of the state of the art in AI by attending conferences, running benchmarks, and conducting systematic literature reviews.

Requirements

• Strong command of Python and DL frameworks (PyTorch preferred; TensorFlow a plus).
• Hands-on experience developing and training deep learning models on GPU/TPU.
• Solid knowledge of applied mathematics: linear algebra, calculus, optimization, statistics, and probability fundamentals.
• Deep understanding of the foundations of modern neural networks and their variants (attention mechanisms, embeddings, convolutions, normalization, regularization techniques, pre-training methods).
• Experience handling large datasets and data augmentation techniques.
• Familiarity with modern infrastructure: distributed environments, parallelized training, mixed precision, model profiling.
• Ability to implement state-of-the-art papers from experimental descriptions and results.
• Strong ability to conduct exploratory research autonomously.
• Critical thinking and scientific rigor in designing and running experiments.
• Clarity in communicating hypotheses, findings, and limitations.
• Collaboration with engineering, science, and product teams.
• Intellectual curiosity and a commitment to continuous learning.
• Resilience in the face of uncertainty and long-cycle experiments.

Nice to have

• Experience with LLMs, generative models (GANs, diffusion), or large-scale multimodal models.
• Prior experience in applied or academic research; publications at top-tier conferences (NeurIPS, ICML, ICLR, CVPR, etc.).
• Experience with research MLOps: MLflow, Weights & Biases, DVC, experiment tracking tools.
• Knowledge of accelerator hardware architectures (CUDA, custom kernels, memory optimization).
• Experience with model compression (pruning, quantization, distillation) and optimized deployment.
• Command of evaluation techniques, metric learning, and scientific benchmarking.
• Familiarity with simulations, advanced mathematical modeling, or computational physics.

Education

• Bachelor's degree in Computer Science, Statistics, Mathematics, Electrical/Computer Engineering, Physics, or related fields.
• A master's degree or PhD in Machine Learning, Artificial Intelligence, Deep Learning, or related fields is strongly preferred.
• Publications and contributions to the AI community are a significant plus.