Make the real decision visible.
Surface assumptions, quality attributes, trust boundaries, dependencies, failure modes, cost, and approvals before teams inherit an architecture by accident.
Co-Founder & CEO · Neuvr
Sunando Mukherjee
I build systems for consequential decisions—where architecture, security, operations, and economics must hold together. At Neuvr, I’m turning that discipline into a product teams can use before expensive decisions become difficult to reverse.
“The hardest infrastructure problem usually begins before the first line of infrastructure code.”
Teams are asked to approve a system before they can test the complete decision. Product intent becomes a diagram, controls become a spreadsheet, cost becomes a calculator, and runtime evidence arrives after commitment.
Neuvr makes the decision itself tangible, testable, and persistent.
01 / FOUNDER FIT
From ambiguity to operating reality.
My work has repeatedly required the same move: connect a business ambition to an architecture, control model, engineering path, and measurable operating outcome.
Carry intent into execution.
Translate approved choices into cloud foundations, platform patterns, policy gates, observability, release controls, and evidence that accountable teams can trust.
02 / SELECTED WORK
Systems that had to hold.
Selected chapters across payments, financial crime, regulated AI, and enterprise modernization. Expand each for the decision behind the technology.
01 Payments · Greenfield cloud platform From market ambition to NPP and BECS capability
Built the AWS cloud foundation for a regulated Australian payments company from the ground up, working with the CISO on controls and security architecture while helping shape the platform engineering path.
The work connected cloud landing-zone design, regulatory controls, operational readiness, and business delivery so the company could take modern NPP and BECS payments capabilities to market.
0→1
Architecture, controls, platform, and go-to-market aligned.
02 Financial crime · Large-scale search platform Unblocking a complex FCC cloud build
Helped unblock a Financial Crime Compliance platform at a major Australian bank by bridging engineering, infrastructure, security, procurement, AWS, Elastic, and legal teams that normally operated independently.
Aligned architecture, licensing and commercial constraints, legal approvals, and delivery responsibilities to move a large-scale OpenSearch implementation from concept to production.
One path
Technical, commercial, legal, and delivery constraints reconciled.
03 Regulated GenAI · Hybrid model platforms Making AI deployment repeatable and governable
Shipped secure GenAI enablement across private and public cloud for regulated workloads, including elastic GPU capacity, hybrid model serving, GitOps release patterns, policy, observability, rollback, and production-readiness controls.
The platform used Run:AI, NVIDIA GPUs, OpenShift, GCP, Azure Arc-enabled AKS, Terraform, and FluxCD while supporting security review expectations across RBAC, least privilege, auditability, and regulated deployment.
~45%
Reduction in model deployment time through repeatable workflows.
04 Agentic delivery · Enterprise modernization AI-assisted execution with quality gates
Built a semi-autonomous delivery framework for cloud and data modernization—connecting intake, architecture, migration scaffolding, notebook generation, data-quality checks, CI/CD, evaluation, defect loops, and human review.
The design treats agentic execution as governed delivery: role-aware context, structured outputs, evaluation loops, shipment controls, and accountable approval points.
5M+
Lines of code modernized through AI-augmented workflows.
03 / HOW I BUILD
Rigor without inertia.
The goal is not to remove necessary scrutiny. It is to remove serial waiting, repeated interpretation, and late discovery.
Start with the decision.
Clarify intent, reversibility, constraints, and the evidence required to proceed.
Model the whole system.
Connect workloads, trust boundaries, failure paths, controls, cost, and operations.
Make controls executable.
Translate policy into deterministic gates, evidence requirements, and accountable exceptions.
Learn from runtime truth.
Compare intended and observed behavior, then evolve the model rather than lose the context.
04 / EXPERIENCE
A builder across layers.
Enterprise AI, cloud and data platforms, payments, financial crime, APIs, governance, and the operating models that connect them.
Selected roles
Principal / Lead Applied AI & Data Architect
Amdocs · Agentic delivery, data modernization, applied AI
Principal Solutions Architect
The Judge Group / Wells Fargo · Governed GenAI platforms
Principal Architect
Stretch 365 + Insight Enterprises · Customer 360, data and AI platforms
Cloud, data, payments & enterprise delivery
Regulated platforms across Australia, India, and the United States