AI governance. Architecture. Engineering.
AI governance for regulated and public-sector institutions in Canada.
Turn AI ambition into defined controls, tested architecture, and a practical delivery plan—with the principal involved from first conversation to implementation.
Authenticate the request. Retrieve only authorized sources.
Constrain actions, data exposure, and tool permissions.
Test grounding. Escalate consequential decisions.
Retain citations, decisions, and an auditable record.
Illustrative control pattern—not a certification or a deployed client system.
A focused starting point
Start with one use case. An AI governance assessment establishes its boundaries, identifies the gaps, and gives your team a decision-ready plan.
Discuss an assessmentThe proposed scope
Duration and fee depend on system access and scope; both are agreed before work starts. The introductory call is free.
Supporting capabilities
When technology decisions span teams, systems, and controls, the answer needs more than a slide deck.
Explore all capabilitiesMove beyond the demo.
Turn standards into decisions.
Own the telemetry path.
Build foundations teams can use.
Connect the difficult parts.
Make the next decision defensible.
Experience behind the consultancy
Selected work by principal Dhruv Doshi in enterprise and startup roles. This is professional experience, not Arihant’s client portfolio.
View all experiencePrincipal’s work at Royal Bank of Canada
A rules-based platform that translates enterprise architecture patterns into service recommendations, solution blueprints, and automated workflows.
Read the experience storyarchitecture patterns mapped into a recommendation platform
Made repeatable architecture decisions available through a platform instead of relying exclusively on manual review.
Principal’s work at Royal Bank of Canada
A vendor-neutral collection and routing architecture designed for approximately 150 TB/day across 10,000+ systems.
Read the experience storyapproximate telemetry volume the architecture was designed for
Made destination routing a configuration decision within a governed collection architecture, reducing dependency on individual observability vendors.
Principal’s work at Royal Bank of Canada
An LLM and retrieval-augmented architecture assistant exploring how natural-language requirements can become grounded recommendations and diagrams.
Read the experience storyretrieval, evaluation, and permission-aware architecture assistance
Development work explores how probabilistic model behavior can operate behind deterministic controls. This is not presented as a completed production deployment.
Principal’s work at HealthCard
The initial product and infrastructure for a healthcare verification platform, spanning applications, APIs, document workflows, and cloud deployment.
Read the experience storyfull-stack product, verification, and cloud foundations
The foundations established from the first commit supported growth to a team of more than 20 engineers, through to acquisition.
Architecture patterns mapped at RBC.
Architecture platformSystems covered by the RBC telemetry architecture.
Architecture scopeAdopted telemetry SDK guidance and onboarding contracts at RBC.
Engineering adoptionPrincipal’s professional experience—not Arihant engagement results or endorsements. Startup experience also includes HealthCard and CryptoVantage, both since acquired.
Working together
Start with the decision or delivery challenge in front of you. Agree the outputs, responsibilities, and acceptance criteria before the work begins.
A consequential decision needs a clearer evidence base.
See the approachThe direction is clear. The implementation path is not.
See the approachA defined workstream needs hands-on engineering.
See the approachYour team needs senior technical capacity.
See the approachProcuring specialist support or building a larger delivery team?
Government, enterprise & subcontractingPersonal accountability
Arihant is led by Dhruv Doshi, an enterprise architect and hands-on engineer whose experience spans banking, healthcare, public services, and research.
From technical strategy to implementation reviews, the work stays connected to the people who will use and operate it.
Meet the principalIdeas for the work ahead
Practical field guide
A practical guide to AI governance: risk-tiered controls, model inventories, audit trails, and the operating cadence that keeps AI systems accountable.
Read the guidePractical field guide
How Canadian public-sector technology procurement works: standing offers, competitive processes, and what vendors need to know before bidding.
Read the guidePractical field guide
How to evaluate LLM applications systematically: eval harnesses, quality metrics, human review, and red-teaming basics for production readiness.
Read the guideBefore we begin
Clear expectations, before a commitment.
Start with a free 30-minute introduction. Share your budget and the decision you need to make; we can discuss whether a focused assessment is a fit. Paid scope, fees, and timing are agreed in advance. No minimum project budget is published.
Tell us your location, collaboration needs, and access restrictions. Remote or on-site arrangements, travel, and any residency requirements are confirmed when scoping the engagement.
Share your NDA and security-review requirements before exchanging confidential information. Terms, access conditions, and procurement prerequisites must be reviewed and agreed before work begins. Please keep initial enquiries non-confidential.
That depends on the use case, available evidence, and stakeholder access. The introduction establishes those constraints; the proposal defines the duration, deliverables, and acceptance criteria before you commit.
No. The experience examples describe Dhruv Doshi’s professional roles. They are not Arihant client engagements, testimonials, or endorsements.
A free 30-minute introduction to discuss your AI governance challenge, constraints, and whether we can help. No project brief required.
Contact us for current availability. Scope, timing, and fees are agreed before paid work begins.