Developer Platform
Technical documentation, API reference, SDKs, CLI tools, and integration guides for deploying GPU-accelerated multimodal AI in clinical production.
$ pip install mergic
$ mergic auth login
$ mergic infer --model vision-v3 --input scan.dcmGetting Started
Quickstart Guide
Install the Mergic CLI, authenticate with your API key, upload a sample DICOM or FASTQ file, and call the inference API. Complete step-by-step tutorials for imaging, genomics, and multimodal workflows with working code samples in Python, Node.js, and cURL.
Start the quickstart →OAuth 2.0, API keys, and service account tokens
Sandbox, staging, and production setup
Multipart, presigned URLs, and bulk ingestion
Submit a job and retrieve results in 5 minutes
API Reference
RESTful API with JSON payloads, OpenAPI 3.0 specs, gRPC support, and sandbox.
Core API
All API requests require Bearer token authentication via OAuth 2.0 or API key. Upload data to presigned S3 URLs or via multipart form-data. Inference endpoints accept modality-specific parameters and return job IDs for async polling. Results are retrieved via GET with pagination and filtering.
Browse full API reference →DICOM upload, model selection, segmentation, and annotation retrieval
FASTQ upload, Parabricks pipelines, VCF retrieval, and annotation
Multimodal summaries via Lumen
Audit logs, RBAC, and policies
Real-time event notifications
SDKs & CLI
Type-annotated client with async/await support, pandas integration, streaming uploads, and automatic retry logic. The most popular SDK for research and clinical data science teams working with imaging and genomics pipelines.
View Python docs →Fully typed client with ESM and CJS support, streaming responses, and webhook verification utilities.
Idiomatic Go client with context support, connection pooling, and generated from OpenAPI spec.
Command-line tool for data upload, pipeline management, job monitoring, and scripted workflows.
Enterprise Java client with Spring Boot integration, connection pooling, and async support.
Code Examples
import mergic
client = mergic.Client(api_key="mgk_...")
# Upload DICOM and run inference
job = client.vision.infer(
model="chest-xray-v3",
input="./patient_scan.dcm",
return_annotations=True
)
# Poll for results
result = job.wait()
print(result.findings)
# [Finding(label="nodule", confidence=0.94, bbox=...)]Upload FASTQ, run Parabricks alignment and variant calling, retrieve annotated VCF.
Combine imaging + genomics + EHR via Lumen API for agentic case summaries.
Query Citadel audit logs by user, action, and time range with cryptographic verification.
Handle inference completion events with signature verification and retry logic.
Product Documentation
Radiology and pathology AI inference with Clara and MONAI. Upload DICOM/WSI, configure detection models, retrieve annotations, and integrate with PACS via DICOMweb. Includes model catalog, fine-tuning guides, and performance benchmarks.
Read Vision docs →Parabricks-backed secondary analysis for whole-genome and exome sequencing with GPU acceleration.
BioNeMo workflows for protein folding, molecular docking, and de novo compound generation.
Agentic summaries merging imaging, genomics, and EHR data
Audit trails, RBAC, de-identification, and model lifecycle
Cross-modal embeddings, cohort search, and federated queries
Integration Guides
Healthcare Systems
Connect Mergic to PACS via DICOMweb for automated study retrieval and structured report push-back. Integrate with LIS for genomics order workflows. Connect EHR systems via FHIR R4 for bidirectional clinical data exchange with reconciliation and deduplication logic.
View healthcare integrations →Direct integration with Clara, MONAI, Parabricks, BioNeMo, NeMo, and NIM.
Connect to S3, Azure Blob, GCS, MinIO, and Hadoop-compatible file systems.
Helm charts, GPU operator, and auto-scaling configurations for K8s clusters.
SSO via SAML 2.0, OpenID Connect, Okta, Azure AD, and Google Workspace.
Deployment Guides
Deploy the full Mergic stack on NVIDIA DGX systems within your data center. Includes Kubernetes manifests, Helm charts, GPU operator setup, air-gapped installation guides, and hardware sizing recommendations for your workload.
Read on-prem guide →Deploy on AWS, Azure, or GCP with GPU instances, managed K8s, and Terraform modules.
Single-tenant VPC with customer-managed encryption keys and private endpoints.
Split inference and training across on-prem and cloud
Auto-scaling policies, GPU orchestration, and load balancing
Prometheus, Grafana, distributed tracing, and alerting
Security & Governance
End-to-end encryption, confidential computing, network isolation, key management, and threat model documentation. Designed for security teams conducting vendor assessments and compliance reviews.
View security docs →HIPAA, SOC 2, GDPR, HITRUST, and FDA 21 CFR Part 11 control mappings.
Immutable ledger, cryptographic signatures, and 7-year retention.
Role-based access, attribute-based policies, and consent enforcement.
Platform Stats
API uptime SLA
Median inference latency
API endpoints documented
Official SDKs (Python, Node, Go, Java)
Pre-built integration connectors
Default API rate limit
Community & Support
Enterprise customers receive dedicated Slack channels with direct access to engineering. Community support is available via GitHub Discussions. Professional services include architecture reviews, custom integration development, and model fine-tuning assistance.
Contact support →Common error codes, debugging workflows, and resolution steps.
Release notes, breaking changes, deprecations, and migration guides.
Real-time platform status, incident history, and maintenance schedules.
GitHub Discussions, sample repos, and community-contributed integrations.
Resources
Upgrade from legacy imaging or genomics platforms to Mergic with zero downtime
Architecture patterns, error handling, and production readiness checklists
Download the complete OpenAPI 3.0 specification for code generation
Try API calls with sample data in a browser-based playground
Changelog
Platform release notes, API changes, and migration guides for all Mergic products.
New vector similarity search endpoints supporting up to 100M embeddings. Batch query support with async responses. Python SDK updated.
GPU-accelerated alignment and variant calling for PacBio HiFi and Oxford Nanopore with Parabricks 4.3 integration.
Multi-institution audit trail federation with cross-site policy inheritance. New RBAC delegation model for consortium deployments.
Real-time AI inference at the scanner edge with NVIDIA Holoscan. Sub-60-second report generation for CT and MRI workflows.
Quarterly
Weekly
Community
Mergic's data ingestion SDK, governance policy engine, and PACS/EHR connector templates are open-source. We welcome contributions from the medical AI community — bug fixes, new format parsers, and governance policy templates are all reviewed and merged on a weekly cadence. Join our GitHub discussions for architecture conversations and RFC feedback.
Request GitHub accessConnect with Mergic engineers and users. Channels for platform development, integration support, and clinical AI governance.
Bi-weekly sessions with Mergic platform engineers. Bring integration questions, optimization challenges, and architecture reviews.
Start building
Talk to our solutions team about API access, enterprise support, and custom integration development for your clinical or research infrastructure.