FEVER by Lowdown Labs: Private In-VPC Image Search and RAG
Lowdown Labs · Software Development
Certification per AWS Marketplace.
Evidence tier Source Confirmed · 4 captures on record
What the publisher says
As described on AWS Marketplace.
Most image search means shipping your photos to someone else's cloud. FEVER changes that. It is a private image-intelligence appliance that runs entirely inside your own AWS VPC, deployed with a single CloudFormation template. Embedding, near-duplicate detection, semantic search, and reranking all happen inside your account. Our efficient models ship directly in the container - zero external calls at inference time and nothing that phones home.
## About Lowdown Labs
Show the rest of the publisher’s description (23 more lines)
Lowdown Labs is a member of the AWS ACE (APN Customer Engagements) program, building purpose-built image intelligence infrastructure for teams that cannot compromise on data privacy. FEVER is powered by a patent-pending Fourier encoder architecture designed from the ground up for cross-resolution visual matching.
## Who It's For
If you custody user photos, product catalogs, scanned documents, or moderation queues, FEVER gives you real image intelligence without ever handing that data to a third party.
## Core Capabilities
- **Kill duplicates:** Cluster near-duplicate images robustly against resizes, re-encodes, and crops with best-in-class cross-resolution tolerance from our patent-pending Fourier encoder.
- **Search everything:** Reverse-image and text search with calibrated match scores, plus an admin view to audit what is actually sitting in your S3 buckets.
- **Feed your agents:** An MCP endpoint (and a plain REST API) let an AI agent run visual RAG scoped to a single downstream customer, with per-customer isolation enforced server-side via a scoped key. Your customer only ever sees their own data.
- **Multi-format, multi-resolution by design:** HEIC, HEIF, WebP, PNG, JPEG, TIFF, PDF. No conversion or resizing required.
## Performance Without Expensive Hardware
FEVER (Fourier Encoder for Visual Embedding and Retrieval) indexes at low resolution and leans on a bespoke, high performance cross-resolution vision model to hold accuracy, so it flies on commodity CPUs without requiring expensive quota upgrades or G-class nodes.
- A single Amazon EC2 c7i node (16 vCPU, no GPU) indexes about 50 images per second.
- A single g5.xlarge (NVIDIA A10G) indexes about 2,800 images per second - roughly 1 million images in about 6 minutes.
- Queries return in millisecond timescales.
- 226 Search Queries / Sec Throughput with an A10G.
- Linear scale-out across a modest CPU fleet when GPU is unavailable or unnecessary. Heterogeneous hardware scaling supported as well.
## Security and Privacy by Architecture
FEVER is designed so that images, vectors, and queries never leave your AWS account. The deployment uses least-privilege IAM roles, runs behind an internal load balancer, and stores vectors in managed Postgres (pgvector) or Aurora within your VPC. You choose the CIDR allowed to reach the service. Data is encrypted in transit and at rest using AWS-managed keys.
## Turnkey Deployment on AWS
FEVER runs on ECS Fargate with managed Postgres (pgvector) or Aurora, optional OpenTelemetry metrics and CloudWatch dashboards, and least-privilege IAM. You pick your compute profile, your VPC, and the CIDR allowed to reach it. Everything else is provisioned for you via CloudFormation.
## Getting Started
Subscribe on AWS Marketplace, deploy the CloudFormation template into your VPC or use the Helm deploy option - load up the admin portal to begin searching your corpus. A deployment runbook, API and MCP reference, and admin resources are included with every subscription.
**Schedule a deployment walkthrough or demo** directly with the Lowdown Labs founder: https://calendar.google.com/appointments/schedules/AcZssZ3eU41AppzWvN6N7Fc5dai8pxZqmrtOHqVvuDiLf3PQorqI5Q5vZ04GPEpEibMlXAUkqfBL6sVC
Private-offer customers receive hands-on onboarding for their first deployment, including a guided walkthrough of the CloudFormation stack, VPC configuration, and initial ingestion setup.
Highlights
Highlighted by the publisher on AWS Marketplace.
Simple and private. Runs entirely in your own AWS VPC. Images, vectors, and model weights never leave your account, with no external calls at inference time. Deploy with one CloudFormation template onto ECS Fargate.
Deduplicate, query, and audit. Finds duplicate and near-duplicate images across resizes, re-encodes, and crops, plus reverse-image and text search and an agent-ready MCP endpoint for visual RAG scoped per customer.
EfficientAI. CPU-native, no GPU required but only recommended: about 50 images per second per 16-vCPU node, scaling linearly across a fleet. Add a GPU node for higher-throughput query and indexing. Any image format works with no conversion.
Preview
2 images
Agent build and provenance
See the full provenance
The layer-by-layer build, the evidence behind each claim, the risk basis and the cross-marketplace links are open to any account. Some rows are disclosed, some the source leaves Unknown; a free account shows you which.
Plans and pricing as listed
1 listed- Units
Refund terms
As stated by the publisher on AWS Marketplace.
If FEVER does not meet your needs, email dev@gimmelowdown.com within 7 days of your initial contract purchase for a full refund of that term. Requests after 7 days are reviewed case by case. FEVER runs in your own AWS account, so the underlying AWS infrastructure charges are billed by AWS and are not refundable by Lowdown Labs.
Sources
Publisher resources
2 linksLinked repositories
Unknown means this listing does not publish a repository. It is not a statement that the code is closed, and a linked repository is not a claim that the publisher wrote it: the registry computes that relationship privately and does not publish it.
Evidence risk is the share of the build you cannot see before you deploy, not a security rating. Sign in to see the layer-by-layer basis for this band.