Skip to content
Trishul Raj

AI Product Manager

Trishul Raj

Product person | Discover, build and deploy with AI | 10+ years of experience on the business side of software.

LinkedInGitHubEmail
Trishul Raj — AI Product Manager, Auckland

Proof

381%First year ARR growthInitiation to production start-up story. First product owner for a robust global trade platform serving exporters, importers, banks, and other ecosystem personas.
1Live AI app, shipped soloNZ Suburb Intelligence — real backend, agent tool-use, evals
10+Enterprise integrations deliveredPayments, banking, customs, air cargo, insurance and port-logistics partners
5+Regulated domainsSupply chain · Export trade · Air/Sea Cargo · Catering · Aviation · Payments

Builds with AI

Meaningful productivity with AI

Not 'AI-curious' — AI applied on both sides of the table, with the work to show for it.

AI at work

Using AI to do 'product' better — AI models for research, analysis, KPIs and metrics, AI in CLI for rapid prototyping, skills for product knowledge, PRDs, user stories, MCPs for work tools, agents for daily summaries and weekly reports.

AI as a passion hobby

Shipping exploratory AI products — demonstrates custom skills and agents, RAG, evals, conversational AI, external data source & API integrations, responsible-AI choices, flexible AI architecture, product management and deployment workflows, built end to end and live.

Work

Portfolio projects

Side-projects and shipped work, each as Problem → Approach → Outcome → Next. Partner names are masked; the reasoning is the point.

AI Skills system for product delivery

Adopted

A library of AI skills that automates the repetitive core of the PO workflow — drafting user stories, acceptance criteria, and backlog tags to a consistent house standard.

  • Applied AI
  • Workflow automation
  • Backlog refinement
  • Skill design
  • Azure DevOps

NZ Suburb Intelligence

Live

A live, natural-language suburb-intelligence and comparison app over New Zealand open-government data (LINZ, Stats NZ Census, Ministry of Education) — ask a real question, get a sourced, reasoned answer with the data visible behind it.

  • Anthropic
  • Gemini
  • Next.js
  • Supabase
  • Vercel
  • RAG
  • Agent tool-use
  • Evals
  • Responsible AI
  • GitHub
  • Linear
Open the live app

Payments-provenance integration

Shipped

with a global card-payments network

Brought verified trade-invoice provenance into a trade-digitisation platform via integration with a global card-payments network's provenance service.

  • Integration
  • Trade finance
  • Fraud reduction
  • Partner delivery
  • API contracts

Bank document-sharing integration

Shipped

with a New Zealand big-four bank

Secure, system-to-system sharing of export documentation between a trade-digitisation platform and a New Zealand big-four bank.

  • Integration
  • Banking
  • Security & consent
  • Cross-org delivery
  • Trade docs
  • mTLS

Self-onboarding for a B2C SaaS

Shipped

Replaced a manual, support-assisted signup with a self-service onboarding flow for a consumer product, taking repetitive setup work off the support team.

  • B2C
  • Onboarding
  • Journey mapping
  • Activation
  • Support deflection

National port-container visibility integration

Shipped

with New Zealand's national port-container visibility platform

A New Zealand-first integration connecting a trade-digitisation platform with the national port-container visibility platform, surfacing container status inside the existing workflow.

  • Integration
  • Port logistics
  • NZ-first
  • Partner delivery
  • Supply-chain visibility

Air-cargo crew catering automation

Shipped

with a global air-cargo carrier

Automated end-to-end catering for an air-cargo carrier's flight crews — turning the partner's flight schedule into compliant, crew-confirmed catering orders, with every pilot and jump-seat crew member allocated the right menu automatically.

  • External integration
  • Aviation
  • Catering automation
  • Data ingestion
  • Mobile notifications

Skills

How it all connects

Not a checklist — a map. Tap a skill to see how I used it, how I'd apply it, and the work that proves it.

Product

4 skills

Owning delivery end to end — from ambiguous ask to shipped, accepted software.

Backlog ownership
Used
Owned the backlog for a multi-module platform — writing the stories, slicing scope, and refining with engineering every sprint.
Would apply
I'd run a tight refinement cadence with a clear definition-of-done, so the team ships predictably instead of guessing.
Acceptance criteria
Used
Standardised Given/When/Then acceptance criteria across a board so 'done' meant the same thing to everyone.
Would apply
I'd make testable criteria the contract between product, engineering and QA — fewer round-trips, fewer surprises at review.
Partner & stakeholder alignment
Used
Aligned banks, a payments network, customs, an air-cargo carrier and port platforms with internal engineering and compliance on contracts everyone could ship.
Would apply
I'd be the connective tissue between external partners and delivery, turning vague asks into criteria all sides sign up to.
Discovery & journey mapping
Used
Mapped activation journeys to find the real drop-off before designing a self-serve onboarding flow.
Would apply
I'd start from the journey and the data, not the feature list — instrument it, find the real drop, fix that.

AI

5 skills

Applied on both sides of the table — AI in the work, and AI as the product.

Applied AI in delivery
Used
Built an AI skills system that drafts stories, acceptance criteria and tags to the team's house standard.
Would apply
I'd find the repetitive, rule-shaped parts of a team's workflow and automate them — keeping judgement human.
Agents & tool-use
Used
Built an agent that turns a natural-language question into a structured query and a cited answer.
Would apply
I'd scope agents to a bounded job with real tools and honest failure states, not an open-ended chatbot.
RAG & embeddings
Used
Embedded data once at ingestion so a single live call answers a query and a provider outage can't take the app down.
Would apply
I'd design retrieval for cost and reliability up front, not bolt it on — and keep embedding models consistent.
Evals & model trade-offs
Used
Scored a fixed question set across models to compare quality, not just speed or price.
Would apply
I'd make evals the basis for model choice and regression-checking, so 'it feels better' becomes a measured claim.
Responsible AI
Used
Framed sensitive data (deprivation) as one input among many — colourblind-safe, never a 'good/bad' verdict.
Would apply
I'd build honesty into the UI — sourced answers, visible confidence, and framing that doesn't stigmatise.

Domain

4 skills

Where I've shipped — regulated, integration-heavy products.

Trade digitisation
Used
Owned product on a platform digitising export documentation and trade-finance workflows.
Would apply
I'd bring real fluency in the messy reality of cross-border trade documents and the parties involved.
Payments & provenance
Used
Delivered an invoice-provenance integration with a global card-payments network to cut financing fraud.
Would apply
I'd treat trust signals as product, not plumbing — make verification something users can see and rely on.
Banking & secure integration
Used
Shipped secure system-to-system document sharing with a big-four bank, including the consent and security model.
Would apply
I'd design integrations security- and compliance-first, so they pass review the first time.
Supply-chain visibility
Used
Delivered an NZ-first integration surfacing port-container status inside the trade workflow.
Would apply
I'd reduce context-switching by bringing the data to where users already work, and be honest when it's stale.

About

A bit about me

I'm a Product Owner who got tired of the gap between “we should use AI” and actually shipping it — so I started building, both inside my delivery work and as products of my own.

By day I own the backlog for an integration-heavy platform, turning partner and regulated requirements into software that ships. On the side I build with AI: an internal skills system that automates the repetitive core of product delivery, and NZ Suburb Intelligence — a live, data-grounded app I took from schema to deploy.

I care about the unglamorous parts — sourced answers over confident guesses, evals over vibes, and responsible framing when data touches people. It's the same instinct whether I'm writing acceptance criteria or shipping an app.

Off the clock

  • The outdoors
  • Riding my motorcycle
  • Building side-projects
  • Good coffee
Trishul Raj outdoors

Contact

Let's talk

Open to contract and permanent product roles. Reach me by email or LinkedIn — happy to share my CV and more details on my work directly.

Connect on LinkedInGitHubEmail me

trishulraj9239@gmail.com