# Dave Remy Solo

- **Event:** [WeaveHacks 3: Self-Improving Agents Hackathon with Weights & Biases](https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80)
- **When:** Jan 31 at 9:00 AM – Feb 1 at 5:00 PM (PST)
- **Where:** 400 California St, San Francisco, CA
- **Team:** [Dave Remy](https://cerebralvalley.ai/u/daveremy99)
- **GitHub:** https://github.com/daveremy/deterministic-memory-layer
- **Demo video:** https://asciinema.org/a/poksmLtZA784MzjQ
- **Gallery:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/50

**DML (Deterministic Memory Layer)** is an MCP server and Claude skill that gives AI agents structured, auditable, event-driven memory. Instead of appending facts to a growing text blob, every memory operation—facts learned, constraints added, decisions made—is recorded as an immutable event with a sequence number. This enables capabilities impossible with traditional memory: deterministic replay to any point in time, full provenance chains showing how facts evolved, counterfactual analysis ("what if this constraint existed earlier?"), and automatic policy enforcement that blocks decisions violating learned constraints. Built during WeaveHacks 3, DML integrates with Weights & Biases Weave for observability, where each memory event maps directly to a tracing span. Unlike traditional memory approaches where state is overwritten and history lost, DML preserves every event—so you can trace from any Weave span back to the underlying memory operations that informed the agent's decisions.

---

Markdown version of https://cerebralvalley.ai/e/weave-hacks-3-self-improving-agents-hackathon-with-weights-and-biases-7014fe80/hackathon/gallery/50. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt