Research And Development Intern

Company Description
:
Engram is building the AI that actually knows you. Our platform combines persistent semantic memory, Bayesian cognitive profiling, and offline reinforcement learning to create an assistant that compounds in value the longer you use it. We're live on iOS and web, backed by the Wharton Innovation Fund, and targeting the 70M neurodivergent knowledge workers under-served by every existing AI tool.

We're a small, fast team. You'll have real ownership over real problems.

The Role:We're looking for a curious, technically grounded intern to join our R&D function. You'll work directly with the founder on the systems that make Engram different from every other AI wrapper: memory architecture, retrieval modeling, and cognitive profiling. This is not a "shadow and document" internship. You'll run experiments, read papers, and push code.

What You'll Work On:

• Literature reviews on memory-augmented LLMs, retrieval-augmented generation, active inference, and continuous learning (papers like Memory-R1, Mem-alpha, Delta Mem are in our reading list)
• Evaluating and benchmarking our retrieval model (the Engram Network, 10-50M parameters) against baseline approaches
• Designing and running ablation experiments on our importance scoring and Bayesian cognitive profiling systems
• Analyzing user retention and memory utility metrics to identify where the system under performs
• Contributing to GraphRAG prototyping: entity extraction, knowledge graph traversal, multi-hop retrieval
• Writing up findings clearly for internal documentation and potential publication

What We're Looking For:

• Currently pursuing a degree in CS, Data Science, Cognitive Science, Neuroscience, Bioengineering, or a related field
• Comfortable reading ML papers and translating findings into experiments
• Working knowledge of Python; experience with PyTorch, Hugging Face, or vector databases (pgvector, Pinecone, Weaviate) is a strong plus
• Familiarity with statistical evaluation: you know what a proper ablation looks like
• Clear written communication; you can explain what you found and why it matters

Compensation:This role is equity-only. We haven't raised yet. That means no salary right now and real upside if we execute. We're transparent about the trade: you get in at the earliest possible moment, with meaningful equity, on a thesis that just received $500M in external validation (Flourish Labs, June 2026). If that framing doesn't work for your situation, this isn't the right role and that's completely fair.

Why Join Now:Engram is the application layer proving brain-inspired memory AI works in production, today, with real users. You’d be joining at the moment the thesis is being validated externally and the product is being hardened internally. Your name goes on the work. Your equity reflects the risk you're taking alongside the founder.