Experiment Memory Agent

An agent remembering runs, plots and decisions across experiments.

Industry	Level	Difficulty	Recommended mode
Quantum computing	4 of 5	Expert	flat

Coming soon

A complete, runnable Experiment Memory Agent reference implementation is in progress. The architecture and core code below already work with SrvDB 0.3.

The problem

• Experiments are repeated unknowingly.
• Plots are not searchable.
• Hybrid workflows are complex.

How SrvDB powers Experiment Memory Agent

1. Run logs and plots are embedded together.
2. The agent recalls similar past runs.
3. Pipeline indexing after each run.
4. Crash-safe.

What changes

Today	With Experiment Memory Agent on SrvDB
Repeated experiments	Recalled results
Unsearchable plots	Searchable
Lost decisions	Remembered

How it works

1. Index each run.
2. Recall similar runs.
3. Summarise.
4. Decide.
5. Record.

SrvDB features it uses

• Several databases: each database is a directory, so different data types or tenants stay separate.
• upsert: inserts new items and replaces changed ones, so the knowledge base stays current.
• Crash-safe persist(): after it returns, data survives a crash or power loss and reopens exactly as committed.
• search_batch: runs many queries at once, in parallel inside Rust.

Use cases

• Run planning
• Debugging
• Hybrid classical-quantum workflows
• Compiler pass notes
• Benchmark tracking
• Reproducibility

Core code

embed() is your embedding model (for example sentence-transformers, or a function that turns your records into feature vectors).

import json
import srvdb

db = srvdb.Database("./experiment-memory-agent-db", dimension=512)

def ingest(items):
    """items: list of (id, text, metadata dict)"""
    db.upsert(
        ids=[i for i, _, _ in items],
        embeddings=[embed(t) for _, t, _ in items],
        metadatas=[json.dumps(m) for _, _, m in items],
    )
    db.persist()                      # crash-safe checkpoint

def similar_runs(summary, k=5):
    hits = db.search(embed(summary), k)
    return [(json.loads(db.get(i)), round(score, 3)) for i, score in hits]

Limits and honest notes

• Plot embeddings need a suitable model.

Caution

SrvDB is classical software: it stores ordinary vectors describing experiments and runs on regular computers, not on quantum hardware.

Build roadmap

1. Prototype: load a sample of real data with flat mode and check that searches return what an expert expects.
2. Measure: if you move to an approximate mode, compare its results with flat on the same queries (recall).
3. Pipeline: add continuous upsert and decide when to call persist().
4. Harden: test restarts and crashes, add logging, review every output path with domain experts.
5. Deploy: ship the database directory with the application or service.

All Quantum computing projects

Level	Project	What it does
1 · Beginner	Lab Notebook Search	Search the group’s notes, papers and run summaries.
2 · Intermediate	Lab Q&A Assistant	Answers questions from the group’s own knowledge with citations.
3 · Advanced	Calibration Drift Detector	Compares calibration snapshots stored as vectors to catch drift.
4 · Expert	Experiment Memory Agent (this page)	An agent remembering runs, plots and decisions across experiments.
5 · Frontier	Circuit Similarity Research	Research platform for circuit embeddings and similarity search.

See also: Use cases · Live examples (coming soon)