Building a Multi-Disciplinary Health and Performance Platform on AWS for Invictus Performance Labs

Industry: Healthcare and Sports Science. Offering: Custom Application Development, Generative AI. Region: ap-south-1 (Mumbai). Delivered August 2026.

About Customer

Invictus Performance Labs is a Bangalore-based performance laboratory serving elite athletes and high-net-worth clients. The practice operates across four disciplines — sports science, physiotherapy and rehabilitation, strength and conditioning, and nutrition — with practitioners in each discipline assessing, prescribing for, and reviewing the same athletes.

Having established the practice model, Invictus set out to build the software that model requires: a single platform carrying the full clinical workflow, the management layer above it, and the athlete relationship alongside it. The brief extended further than internal tooling. Invictus had identified that other premier performance centres operate the same way and face the same requirement, so the platform was to be built as something the practice could licence, not only something it could use.

Business Objectives

1. Establish one athlete record across all four disciplines. Sports science, physiotherapy, strength and conditioning, and nutrition each hold part of the picture. Invictus wanted practitioners working from shared context, so that a clinical decision in one discipline is visible to the others as it is made.

2. Put performance data at the point of decision. Assessment outputs, force plate and VO2max results, readiness and load figures, and wearable data are only useful to a practitioner while they are current. The objective was to have them structured, live, and in front of the person prescribing.

3. Give multidisciplinary collaboration a system of record. Handovers, referrals, case reviews, physiotherapy holds, and return-to-play clearances are consequential decisions. Invictus wanted them tracked, owned, and retrievable.

4. Equip discipline leads and operations with their own consoles. Caseload oversight, approvals, quality review, and reporting for clinical leads; staff leave, rostering, approvals, and analytics for operations.

5. Connect the commercial side to the clinical side. Lead capture, pipeline and conversion tracking, follow-ups, member management, and invoicing, working from the same member records as the practitioners.

6. Open a direct channel to athletes. A native mobile application for session execution, fuel logging, recovery tracking, progress, check-in, and messaging with the care team, with wearable data flowing in automatically.

7. Build it multi-tenant from the outset. The platform had to be capable of being licensed to other performance centres without a subsequent rebuild, which meant tenancy as a first architectural decision rather than a later migration.

Solution Delivered

1. AspireNXT built Invictus Health OS as a single modular monolith on AWS, covering approximately 140 screens and 60 end-to-end workflows across 10 functional domains.

2. Two compute services, not a microservice estate. One API service carries all domain modules in process; one background worker consumes BullMQ jobs. Both run on Amazon ECS Fargate. This keeps the running footprint and the operational burden proportionate to a practice that does not employ a platform team.

3. One PostgreSQL database, extended rather than fragmented. Amazon RDS for PostgreSQL holds the full domain model, extended with TimescaleDB for wearable time-series and pgvector for AI embeddings and semantic search. Amazon ElastiCache for Redis provides the session store, job queue backend, and hot-path read cache.

4. A shared athlete record across all four disciplines. Sports science assessments and test batteries, physiotherapy injury files and body-region charting, strength and conditioning programmes and ACWR load tracking, and nutrition plans with IFCT-based macro tracking all write to the same athlete profile. A physiotherapy hold is visible to the strength and conditioning lead before session assignment.

5. Practice management built into the same system. Per-discipline lead consoles provide caseload oversight, approvals, quality review, and reporting. An operations console covers staff leave, rostering, approvals, and analytics. A customer service and sales module handles lead capture, pipeline and conversion tracking, follow-ups, member management, and invoicing.

6. A native athlete app with wearable integration. React Native for iOS and Android, covering session execution, fuel logging, recovery and bio tracking, progress, check-in, and coach messaging. Apple HealthKit and Garmin Connect sync into the same record the practitioners read from.

7. Deployed in ap-south-1 so athlete health data stays in India. Amazon Cognito handles practitioner and athlete authentication with role-based access control and MFA. AWS KMS manages encryption keys and AWS Secrets Manager holds credentials. AWS CloudTrail and Amazon CloudWatch maintain a tamper-evident audit trail of every practitioner and administrative action.

8. Engineered white-label from the first architectural decision. Multi-tenancy is a property of the data model and the deployment. The platform Invictus commissioned for its own practice is the platform it can licence to other performance centres.

Anthropic Claude on Amazon Bedrock

Claude is the reasoning layer of the platform. It runs on Amazon Bedrock in the Mumbai region, so athlete health data never leaves India to reach a model.

1. Composite scoring across disciplines. An athlete generates readings that no single number summarises: force plate output, VO2max, readiness and load figures, subjective check-ins, and wearable data from Apple Health and Garmin. Claude reasons across those inputs together to produce a composite score a practitioner can interpret, rather than a formula that collapses them into an average.

2. The cross-discipline intelligence layer. Claude reads across an athlete history — sports science assessments, physiotherapy injury files, strength and conditioning load, nutrition adherence — and surfaces patterns that sit between disciplines, where a single practitioner reviewing their own records would not see them.

3. Structuring uploaded medical documents. Amazon Textract extracts raw text from scans and reports. Claude turns that text into structured entries on the athlete medical summary, so an uploaded document becomes queryable record rather than an attachment nobody opens.

4. Semantic search over the athlete record. Embeddings stored in PostgreSQL via pgvector let practitioners find relevant history by meaning rather than keyword, with Claude interpreting the query and the retrieved context.

5. Routed to the model the task needs. Scoring a single assessment and reasoning across an athlete full history are different orders of work. Every request is routed to the Claude model matched to the reasoning complexity and capability the task requires, through Amazon Bedrock Prompt Router, so neither task is served by a model sized for the other.

6. Governed on every call. Amazon Bedrock Guardrails apply PII controls and content safety to every request and response. Inference calls are logged to the same audit trail as practitioner actions.

Services Used

• Anthropic Claude on Amazon Bedrock

• Amazon Bedrock Prompt Router and Bedrock Guardrails

• Amazon ECS Fargate

• Amazon RDS for PostgreSQL (TimescaleDB, pgvector)

• Amazon ElastiCache for Redis

• Amazon Textract

• Amazon Cognito

• AWS Lambda and AWS Step Functions

• Application Load Balancer, Amazon CloudFront, AWS WAF

• Amazon S3, Amazon SNS, Amazon SES

• AWS KMS and AWS Secrets Manager

• Amazon CloudWatch, Amazon GuardDuty, AWS CloudTrail, AWS Config, Amazon Inspector

Delivery

AspireNXT delivered the platform between January and August 2026.

What Was Delivered

1. One athlete record across four disciplines, covering assessments, diagnostics, injuries, medical summary, performance, and recovery.

2. Management consoles for clinical leads and operations, with caseload, quality, approvals, staffing, and roster visibility built into the platform.

3. A connected commercial pipeline, from lead capture through conversion tracking to invoicing, working from shared member records.

4. A direct athlete channel through the native mobile app, with automatic Apple Health and Garmin sync.

5. DPDP-aligned architecture. India data residency, role-based access control, encryption at rest and in transit, and a complete audit trail. AspireNXT implemented the technical controls; independent audit and VAPT are commissioned separately by the customer, and AspireNXT answers auditor questions and remediates findings.

6. A licensable product. Multi-tenant and white-label by design, supporting a B2B licensing route to other premier performance centres.

7. A run cost proportionate to the practice. The two-service, single-database design keeps the single-centre production footprint in the low hundreds of dollars a month on AWS, billed on actuals. AspireNXT does not mark up infrastructure.

Planning a platform build for a clinical or performance practice? Talk to the AspireNXT applications team.