Research Engineer - Agent Memory

About the Role

We're building the memory layer for AI agents, enabling long-term memory that allows AI to remember conversations, learn from interactions, and build real context over time. This role is crucial in developing the models that decide what an AI remembers, forgets, and learns. You'll take ideas from research papers to production-ready A/B tests within weeks, working with millions of real interactions to drive innovation.

Responsibilities

• Train models for memory extraction, updates, consolidation/forgetting, and conflict resolution, iterating quickly on real data and outcomes.
• Turn research ideas into working prototypes, benchmark them against baselines, and productionize successful models.
• Build evaluation frameworks, including automated metrics, gold sets, online A/B tests, and dashboards.
• Identify customer pain points, formulate testable hypotheses, and validate solutions through field trials.
• Collaborate with Engineering to design APIs, plan rollouts, and ensure SOTA latency, reliability, and cost.

Requirements

• Experience in RAG or information retrieval (retrieval, ranking, query understanding) for real products.
• Model training/fine-tuning experience with LLMs/encoders and strong experimental design skills.
• Proficiency in Python, deep experience with PyTorch, and familiarity with vLLM and modern serving frameworks.
• Experience building evaluation for complex language and/or retrieval tasks (gold sets, offline metrics, online tests).
• Ability to orchestrate data pipelines for low-latency production model deployment (batch + streaming).
• Clear communication skills with stakeholders across engineering, product, GTM, and customers.

Qualifications

• Publications at top-tier venues like NeurIPS, ICML, ACL, etc. (Nice to have)
• Experience with privacy-preserving ML (redaction, differential privacy, data governance) (Nice to have)
• Familiarity with memory/retrieval literature or prior work on memory systems (Nice to have)
• Expertise with embeddings, vector-DB internals, deduplication, and contradiction detection (Nice to have)

Benefits

• Health, dental, and vision coverage: Comprehensive plans, fully covered for you (and subsidized for dependents)
• Daily catered lunch and dinner in-office
• Flexible PTO
• Equity in an early-stage company
• Regular team happy hours and events
• Top-tier equipment and setup

Compensation Range: $175K - $250K