You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
design(atlas): secret-tier local embedder — model selection and fleet-hardware evaluation (secret text never leaves the fleet) #6059
One cost argument overrides price: the secret tier must be embedded locally. Sending secret-tier text to a hosted embedding API is a classification violation regardless of the bill. A local embedder (BGE-M3 class, or the fleet's existing local models) at bursty utilization costs more per token than the API's $0.001–0.01 and is worth it; local marginal cost is electricity, not metered API, which also caps the plan-spend risk structurally. The internal tier can use whichever route the cost ledger (sibling under #6019) favors.
Throughput on fleet hardware at realistic batch sizes; batching shape for delta-sized workloads (an always-on GPU for bursty embedding is a named dead end — ~$1,080/mo spot A100 at zero usage vs pennies of batch API).
Operational home: which machine, which service shape, failure posture when the embedder is down (ingest queues; never silently reroutes secret text to an API).
Done looks like
A selection + evaluation design with the no-egress rule stated as a hard invariant.
Resume evidence: recorded worktree /home/user/git/fleet/task-10425, branch queue/10425-6059-secret-tier-local-embedder, clean at 2a676206852c9b5f4aa5a61873411fa850b13ec4 (19 commits behind current origin/main; no task edits found). No matching PR; issue OPEN with no hard-stop labels.
Verdict:#6059 remains OPEN and is not already handled; no matching PR exists. It is a design ticket, not an implementation fix, so no code changes were made. Revalidated at current main00a294ab6a38b23b3a3f855ff5e6102d5aed78c5.
Current baseline:pkgs/atlas-search/crates/atlas-embed already provides local ONNX inference for pinned Snowflake/snowflake-arctic-embed-m-v2.0; fetch-model provisions the model artifact into local state. #5351 Gate A is closed PASS for bit-identical parity with the existing corpus pipeline. That proves a local candidate exists, but it is not the secret-tier model decision, fleet-hardware throughput/RSS evaluation, or a no-egress policy for ingestion. Model download/provisioning and secret-text inference are separate concerns.
Route/order: proceed with the design using the existing Snowflake model as a benchmark baseline; compare candidate quality on the planned versioned golden set (#6046 under #6027) and measure realistic batch throughput/RSS on fleet hardware. Preserve the absolute invariant from #6019: secret-tier text never goes to a hosted API, including on local-model failure (queue/fail closed; no silent fallback). Coordinate the model choice with #6060’s open one-model-vs-normalization contract before selecting a distinct model for secret tier, and with #6021’s classification boundary. These are design coordination points, not blockers to writing the design.
No user decision is required to triage. The pre-existing worktree /home/user/git/fleet/task-10425 remains clean and unchanged on queue/10425-6059-secret-tier-local-embedder (based on 2a676206; 19 commits behind current main); it is not needed for this design-only task.
Agent: queue worker (repo-worker-code), lane c8b9355e-319a-4640-8096-7ad19480d151
Found via: atomic queue claim; resumed prior checkpoint
Activity
developing-today commentedon Nov 5, 2026
CODE-1724
developing-today commentedon Nov 5, 2026
🤖 Agent note · state: handed off
5 remaining items
developing-today commentedon Nov 5, 2026
🤖 Agent note · handed off
dezren39 commentedon Nov 5, 2026
🤖 Agent note · state: picked up
repo-worker-code), lanec8b9355e-319a-4640-8096-7ad19480d151/home/user/git/fleet/task-10425, branchqueue/10425-6059-secret-tier-local-embedder, clean at2a676206852c9b5f4aa5a61873411fa850b13ec4(19 commits behind currentorigin/main; no task edits found). No matching PR; issue OPEN with no hard-stop labels.dezren39 commentedon Nov 5, 2026
🤖 Agent note · triage complete (task 10425)
Verdict: #6059 remains OPEN and is not already handled; no matching PR exists. It is a design ticket, not an implementation fix, so no code changes were made. Revalidated at current
main00a294ab6a38b23b3a3f855ff5e6102d5aed78c5.Current baseline:
pkgs/atlas-search/crates/atlas-embedalready provides local ONNX inference for pinnedSnowflake/snowflake-arctic-embed-m-v2.0;fetch-modelprovisions the model artifact into local state. #5351 Gate A is closed PASS for bit-identical parity with the existing corpus pipeline. That proves a local candidate exists, but it is not the secret-tier model decision, fleet-hardware throughput/RSS evaluation, or a no-egress policy for ingestion. Model download/provisioning and secret-text inference are separate concerns.Route/order: proceed with the design using the existing Snowflake model as a benchmark baseline; compare candidate quality on the planned versioned golden set (#6046 under #6027) and measure realistic batch throughput/RSS on fleet hardware. Preserve the absolute invariant from #6019: secret-tier text never goes to a hosted API, including on local-model failure (queue/fail closed; no silent fallback). Coordinate the model choice with #6060’s open one-model-vs-normalization contract before selecting a distinct model for secret tier, and with #6021’s classification boundary. These are design coordination points, not blockers to writing the design.
No user decision is required to triage. The pre-existing worktree
/home/user/git/fleet/task-10425remains clean and unchanged onqueue/10425-6059-secret-tier-local-embedder(based on2a676206; 19 commits behind current main); it is not needed for this design-only task.repo-worker-code), lanec8b9355e-319a-4640-8096-7ad19480d151