Memories.ai Research Fellowship

The Memories.ai Research Fellowship is an elite research program operated by Memories.ai, a organization building universal long-term memory systems, stateful agent architectures, and multimodal context retrieval for AI models.

The fellowship gives AI researchers, PhD candidates, and ML engineers direct access to dedicated compute clusters, proprietary long-context datasets, and technical mentorship to solve fundamental challenges in persistent AI memory and continual learning.

• Competitive Financial Compensation: Direct research stipend and project funding (competitive with top AI research lab fellowships).

• Dedicated Compute Infrastructure: Direct allocation of high-performance GPU clusters (NVIDIA H100/A100 instances) for large-scale training and benchmarks.

• Mentorship & Publication Track: 1-on-1 collaboration with lead AI researchers at Memories.ai, with guaranteed support to publish open-source benchmarks and co-author papers for premier conferences (NeurIPS, ICML, ICLR, CVPR).

• Location & Format: Flexible options including fully remote (global participation) or in-person / hybrid residency at their primary engineering hubs.

Fellows undertake high-impact technical projects centered around long-term memory architectures:

1. Stateful Agent Architectures & Lifelong Learning: Models that retain, update, and integrate episodic/semantic memories over long horizons without catastrophic forgetting.

2. Multimodal Context Retrieval & Compression: Efficient vector search, hierarchical context summarization, and retrieval-augmented generation (RAG) across video, audio, and code.

3. Graph-Based & Associative Memory Systems: Graph neural networks (GNNs) and dynamic knowledge bases optimized for real-time model inference.

4. Privacy-Preserving & Encrypted Memory: Zero-knowledge and local-first memory layers for personal and enterprise AI agents.

• Candidate Profiles: PhD students, postdoctoral fellows, independent ML researchers, or senior engineers with a strong technical track record in machine learning or natural language processing.

• Key Qualifications:

  • Strong research background in transformer architectures, vector databases, RAG, reinforcement learning, or agentic frameworks.

  • Proficiency with PyTorch, CUDA, distributed training frameworks, and vector search systems (e.g., FAISS, Qdrant).

  • History of peer-reviewed publications or high-star open-source ML contributions.

• Selection Model: Rolling application process evaluated quarterly.

• Submission Materials: CV/Google Scholar link, GitHub profile, and a concise technical research proposal (1–2 pages) outlining your proposed approach to persistent AI memory.