Abide Robotics

Abide Robotics is building a 24/7 robotic companion for older adults — filling the gaps left by an understaffed workforce and providing a constant ease of loneliness, one of the leading causes of mental decline.I lead engineering and product at Abide: I architected and built the robot's core AI system — the "brain" — and lead a four-engineer team across software and hardware. The system is built for a HIPAA-compliant environment, with a five-resident pilot launching in an assisted living facility.

What I Built

The system uses a 6-layer memory retrieval pipeline to provide meaningful, personalized interactions with residents:

• Immediate memory — recent conversation context (TimescaleDB)
• Global semantic memory — cross-conversation vector search (Qdrant)
• Local semantic memory — within-conversation context (Qdrant)
• Knowledge graph — causal chains and entity relationships (Neo4j)
• Facts & observations — extracted, persistent knowledge about each resident
• User profile — preferences, routines, and personal history

This lets the robot remember residents, track their preferences, and have genuinely contextual conversations over time — not just stateless Q&A.