Portfolio
Selected Work
Products, platforms and AI systems built across different stages of my career.
Featured
Aarohi AI / Product Innovation Lab
Applied AI/Product Innovation Platform (in build)
PWAAI Product StrategyEdTech0-to-1
Explore The Lab →- Overview
- Designing and building a connected six-tool PWA that turns a live 3-day product-development workshop (Think → Discover → Define → Decide → Design → Defend) into a structured, AI-assisted experience for MBA and Engineering student teams.
- Problem
- Product education is taught as disconnected modules and generic AI chat. Teams re-explain their project to every tool, and nothing they produce accumulates into a defensible product story.
- Approach
- Opinionated: the tools challenge weak product reasoning instead of agreeing with it.
- Connected: six tools share one live project context, so nothing is re-explained.
- Applied: every stage produces a tangible output a team can defend.
- Designed to be validated live in workshops before anything becomes a shipped product.
- Validation
- A three-day workshop format designed to be validated in the room before becoming a product.
The six-stage journey
ThinkDiscoverDefineDecideDesignDefend
WanderWell
Multi-Agent GenAI Product
GenAIAgentic AIProduct Strategy0-to-1
- Problem
- Planning a wellness-oriented trip means reconciling health goals, constraints, budget, timing and destination options across scattered sources. The work is research-heavy and repetitive, and generic travel tools optimise for booking rather than for the traveller's intent.
- Context
- A self-directed 0-to-1 GenAI product built end to end — from discovery through prototype — to work through what agentic product design actually demands beyond a chat interface.
- My Role
- Product discovery, problem framing, JTBD and personas, PRD authorship, multi-agent architecture design, prototype build and roadmap.
- Approach
- Product discovery to frame the traveller's real problem before considering AI.
- Jobs-to-be-Done framing to separate the functional job from the emotional one.
- Personas grounded in distinct wellness intents and constraints.
- A PRD defining scope, non-goals, flows and success signals.
- Prototyping in Figma, then a working build in Lovable.
- Solution
- A multi-agent GenAI experience where specialised agents handle discovery, itinerary construction and wellness-fit reasoning, coordinated toward a single traveller goal, with the user confirming direction at defined checkpoints rather than at every step.
- Key Product Decisions
- Define the job first; the agent architecture follows the job, not the reverse.
- Split agents by responsibility so each has a narrow, testable remit.
- Place human-in-the-loop checkpoints where the cost of a wrong assumption is highest.
- Keep the interface intent-led rather than chat-led.
- Sequence the roadmap so the core planning loop proves itself before breadth.
- Technology / AI
- GenAI reasoning with multi-agent orchestration, tool use, structured outputs and explicit human review points. Figma for interaction design; Lovable for the working prototype.
- Outcome / Learning
- The hard part of agentic products is not model capability — it is scoping each agent's responsibility, deciding where human judgement belongs, and keeping the experience legible to the user.
Discovery to prototype
DiscoveryJTBD & PersonasPRDMulti-Agent DesignPrototypeRoadmap
AIQod
Enterprise Automation Product
0-to-1B2B SaaSRPAProduct Management
- Problem
- Enterprises run large volumes of document-driven, rule-bound work across systems that were never designed to talk to each other. Point solutions automate a step; they rarely automate the process end to end.
- Context
- A founding-stage environment: I joined as the second employee and founding Product Manager, with no existing product, process or playbook.
- My Role
- Founding Product Manager — product scope and boundaries, requirements, prioritisation, and enterprise client capabilities alongside engineering and early customers.
- Approach
- Worked directly with early enterprise clients to ground the product in real processes.
- Defined the product's boundaries explicitly — what the platform owns versus what it integrates with.
- Made deliberate build / buy / partner decisions given a very small team.
- Prioritised capabilities that unblocked deployment in enterprise environments.
- Solution
- An IDP + RPA platform combining intelligent document processing with process automation, built from zero, with the capability set enterprise clients needed to adopt it inside existing operations.
- Key Product Decisions
- Treat document understanding and process execution as one product surface, not two tools.
- Buy or partner for commodity components; build where the differentiation was.
- Design for exceptions and human review from the start, not as a later add-on.
- Say no to bespoke requests that would have fractured the platform.
- Technology / AI
- Intelligent document processing, RPA orchestration and enterprise integration patterns, with human review paths for low-confidence extractions.
- Outcome / Learning
- At founding stage, product management is mostly boundary-setting. Scope discipline and honest build/buy/partner calls decide whether a small team ships something enterprises can run on.
Platform shape
IngestDocument UnderstandingException & ReviewProcess AutomationEnterprise Integration
Marsh
Enterprise Automation
EnterpriseAutomationProduct StrategyROI
- Problem
- Large organisations carry manual, fragmented operational processes whose cost is spread thinly across many teams — which makes the problem easy to tolerate and hard to prioritise.
- Context
- Enterprise automation initiatives inside a global organisation, with multiple stakeholder groups, established governance and existing systems of record.
- My Role
- Own product lifecycle activities for enterprise automation initiatives — from problem discovery and business cases through PRDs, roadmap and deployment.
- Approach
- Discovery with process owners and operational teams to establish the real workflow, not the documented one.
- Business cases that make the cost of the current state and the value of change explicit.
- PRDs that define scope, controls, exceptions and acceptance criteria.
- Roadmap sequencing across dependencies and stakeholder readiness.
- Deployment with governance, change management and post-release review.
- Solution
- Automation initiatives delivered through a consistent product lifecycle, so each one is justified by a business case, specified clearly, sequenced sensibly and governed after release.
- Key Product Decisions
- Prioritise by business case strength and readiness, not by request volume.
- Design the exception path as carefully as the happy path.
- Treat governance and controls as product requirements.
- Define measurement before deployment so value can be assessed honestly.
- Technology / AI
- Enterprise automation tooling and integrations within existing systems and controls, with AI applied where it improves the workflow rather than where it is available.
- Outcome / Learning
- In an enterprise, adoption is the product problem. Discovery quality and stakeholder alignment determine whether an automation is used after launch.
Lifecycle
DiscoveryBusiness CasePRDRoadmapDeploymentGovernance