Product Owner × Engineer · Saskatoon, SK, Canada

Arshaq Hisham

Product judgment with real engineering depth.

A decade building systems where software meets high-stakes domains — ERP consulting at KPMG, a Nordic EHR at Cambio, a multi-tenant search platform at Villvay, and now hands-on LLM agent engineering.

10 yrs
consulting, healthcare & platforms
2
production platforms shipped
~81k
lines of TypeScript in production
37
open notes in the knowledge graph

For future colleagues

Where I fit in a workplace

The shortest version: I sit in the gap between product and engineering — and I close it, in both directions.

The bridge

Roadmaps and ROI reviews on Monday; Elasticsearch DSL and tRPC on Tuesday. I translate business intent into technical decisions — and technical trade-offs back into business language — without losing anything in between.

High-stakes domains

I'm at my best where mistakes are expensive: a clinical EHR serving Swedish healthcare, pre-launch QA for banking apps, national-scale ERP programs, industrial B2B commerce. Compliance-first thinking is my default, not an afterthought.

Zero → production

I take products from blank repo to live customers. ELUX Travels — a multi-region travel booking platform — was designed, built and shipped solo: 460+ commits over 11 months, running in production today.

How I work

Requirements → design → build → measure, with artifacts at every step: models and specs stakeholders actually read, code that ships, dashboards that prove what happened. Evidence over claims, always.

Roles where I land: Product Owner / PM — platforms, search & AI AI Engineer — multi-agent systems AI × Healthcare — clinical informatics Technical Product Owner — search & data

Career arc

A decade, one direction

Each step added a layer: process rigor, clinical depth, platform scale — now the AI systems themselves.

  1. 2015

    BSc Business Information Technology

    Staffordshire University

  2. 2018

    MSc Advanced Software Engineering — batch topper

    University of Westminster, UK

  3. 2016 – 2018

    KPMG · Trainee BA → Business Analyst → Associate Consultant

    ERP and process re-engineering consulting across seven sectors. Flagship: an AS-IS study of the Ceylon Electricity Board — 17 departments, ~150 processes — feeding TO-BE designs for procurement, inventory, asset and fuel management.

    ConsultingERPQABankingGovernment

  4. 2019 – 2022

    Cambio Software Engineering · Lead Business Analyst

    Four years owning business analysis for the Overview module of COSMIC — one of the Nordic region's leading electronic health records, serving regional healthcare organizations across Sweden. HL7, FHIR, openEHR, SNOMED CT and GDPR as daily standards.

    Healthcare ITEHRInteroperabilityProduct

  5. 2022 – present

    Villvay Systems · Product Owner

    Product and engineering lead on a multi-tenant B2B search platform for Wurth industrial supply companies — built end to end in one Turborepo: Airflow ETL, Elasticsearch relevance, Cloudflare Workers edge APIs, Next.js admin console.

    SearchDataEdgeSaaS

  6. 2024 – 2025

    ELUX Travels · Founder-built side project

    Production multi-region travel booking platform, shipped solo and live at eluxtravels.uk — the zero-to-one proof behind the platform work.

    TravelSolo buildProduction

  7. 2026

    Applied AI · AI App Generator

    Multi-agent system that writes, executes and iterates on full Next.js apps inside cloud sandboxes with live previews — Mastra, E2B, Vercel AI SDK v5.

    LLM agentsSandboxingStreaming UX

Education: MSc Advanced Software Engineering, University of Westminster (UK), 2018 — batch topper · BSc Business Information Technology, Staffordshire University, 2015

Experience

Three rooms, one through-line

Consulting discipline, clinical product depth, platform-scale building.

Villvay Systems

Product Owner · Multi-tenant B2B search for Wurth companies

Dec 2022 – present
  • Built the platform end to end: Airflow ETL, Elasticsearch on Elastic Cloud, Cloudflare Workers edge APIs, Next.js admin console — ~35k lines of TypeScript and Python in one Turborepo.
  • Translated complex system flows into roadmaps and ROI-focused reviews for global stakeholders; led requirement elicitation, TO-BE architecture, prototypes and delivery timelines.
  • Ran competitive and technology analysis across headless commerce, search and graph databases; owned quality metrics, Power BI dashboards and PIM data integrity.
  • Earlier AI R&D: OpenAI and Hugging Face integrations for data-processing automation, plus POCs with v0, bolt.new and Claude Code for client feedback loops.

Cambio Software Engineering

Lead Business Analyst · COSMIC EHR, Swedish market

Apr 2019 – Dec 2022
  • Owned business analysis for the Overview module — consolidated patient information, care documentation, charting and operational insights — for a leading Nordic EHR.
  • Worked with physicians, nurses, administrators, architects and developers across patient management, medication management, referrals, scheduling and clinical administration.
  • Specified system logic, information structures and API designs to HL7, FHIR, openEHR, SNOMED CT and GDPR standards, API-first.
  • Ran Design Sprints and human-centered design; produced UML models, Figma prototypes, acceptance criteria and managed backlog, UAT and release readiness.

KPMG

Trainee Business Analyst → Business Analyst → Associate Consultant

Jan 2016 – Oct 2018
  • ERP and business process re-engineering consulting across government, utilities, banking and finance, insurance, manufacturing, retail/FMCG and telecom.
  • Ceylon Electricity Board ERP program: AS-IS study across 17 departments and ~150 processes; FRS workshops; TO-BE designs for procurement, inventory, asset and fuel modules.
  • End-to-end ERP and core banking vendor evaluations — SAP, Oracle, Microsoft, Ebao, Misys, Temenos — from tenders and POC demos to recommendation reports.
  • Data migration QA and pre-launch assurance for banking and insurance clients (SQL, Excel, IDEA) and for internet/mobile banking apps, a mobile wallet and a government e-Registrar.

Flagship work

Projects

Numbers below are taken straight from the codebases — nothing inflated, nothing vibes-based.

● Production at Villvay

Villvay Search Platform

Multi-tenant B2B e-commerce search for Wurth industrial suppliers — ~35k lines of TypeScript and Python, 5 tenants isolated end to end.

  • 16-layer relevance cascade from exact SKU to fuzzy, with tier suppression at result thresholds.
  • Weighted Reciprocal Rank Fusion across 11 retrievers; ELSER sparse-vector semantic fallback computed at ingest, so query time pays zero inference.
  • E-commerce linguistics: unit-of-measure synonyms, fraction/dimension normalization, per-account part numbers, barcode lookup.
  • Zero-code tenant onboarding: config-only, 11 routes and 10 search collections per tenant.
  • Airflow pipelines — one templated DAG generating per-tenant DAGs — syncing on 4-hour schedules with watermark incrementals.
  • Config platform: 33 tRPC procedures, draft → preview → promote, one-click rollback.

ElasticsearchAirflowCloudflare WorkersNext.jstRPCTypeScriptPython

● Production eluxtravels.uk ↗

ELUX Travels

Multi-region travel booking platform — tours, hotels, flights, cruises — designed, built and shipped solo.

  • ~46k lines of hand-written TypeScript, 23 routes, 22 CMS collections, 42 server actions, 460+ commits over an 11-month build.
  • Next.js + Payload CMS 3 running as Cloudflare Workers via OpenNext, with edge ISR (R2 incremental cache, D1 tag cache, Durable Object queue).
  • Block-based quote & itinerary builder producing branded quote documents with their own preview site; deposit and installment payment plans with proof-of-payment verification.
  • TLD-driven multi-region: 5 currencies from one codebase, no client-side currency flashing; Google & Apple Sign-In, 5-role RBAC.
  • Production data migrated from Neon Postgres to Cloudflare D1 via chunked replay scripts.

Next.jsPayload CMSCloudflare WorkersD1R2TypeScript

● Personal · 2026

AI App Generator

Describe an app in chat; a multi-agent system writes, executes and iterates on full Next.js apps inside cloud sandboxes with live previews.

  • Mastra multi-agent pipeline: coding, title and response agents in a typed 4-step workflow with Zod-validated step contracts.
  • E2B sandbox lifecycle management — sandboxes survive across turns; projects restored from Postgres into fresh sandboxes on expiry, on a custom Docker template.
  • Long-term agent memory with semantic recall (FastEmbed embeddings in a LibSQL vector store) and AI tracing via Mastra observability.
  • SaaS platform: tRPC end-to-end types, Drizzle on PostgreSQL, Better Auth, Polar metered billing, per-user rate limits.
  • Streaming UX on Vercel AI SDK v5: streaming markdown, Shiki code views, live sandbox preview panel.

MastraE2BVercel AI SDKtRPCDrizzleTypeScript

● Design study

AI Oncology Capstone

AI-assisted breast cancer early-detection and diagnostics module designed for the COSMIC EHR, in the Sri Lankan context.

  • Full research → synthesis → ideation → POC design sprint, aligned to Cambio's roadmap and the EU Ethics Guidelines for Trustworthy AI.
  • Three concepts evaluated through a weighted Harris Profile; selected a CNN clinician-assistant concept over a patient app and a risk model.
  • Measurable AI requirements as design targets: 80%+ detection accuracy, 60% faster lab-investigation selection, ≤5 minutes screen time per consultation.
  • Clinician-in-the-loop second opinions designed to discourage over-requisitioning invasive tests; anonymized scan-image repository proposed for AI development.

Healthcare AIResponsible AIClinical workflowsDesign

● Prototype

webcules-init & adjacent experiments

The earlier iteration of the AI app-generation platform — prototyped on the Vercel AI SDK with Supabase persistence and component-library-aware generation — later rebuilt into the AI App Generator on Mastra + E2B to add agent memory, sandboxing and orchestration. Alongside it: tru, a workflow-automation experiment on Cloudflare's edge stack, and webcules-ai-platform, an intent-to-execution agentic automation concept.

Vercel AI SDKSupabaseCloudflare

Capabilities

Skills & domains

AI & LLM Engineering

Multi-agent orchestration, tool calling with strict schemas, sandboxed execution, agent memory, streaming AI UX.

Search & Relevance

Layered relevance cascades, RRF fusion, ELSER semantic search at ingest, e-commerce linguistics.

Full-Stack Development

TypeScript end to end — Next.js, React, tRPC, Payload CMS, Drizzle, PostgreSQL, D1.

Data Engineering

Airflow DAGs, watermark incremental sync, declarative transformations, production migrations.

Cloud & Edge

Cloudflare Workers, R2, D1, Queues, Durable Objects; OpenNext edge ISR; Docker; GitHub Actions.

Product & Business Analysis

Roadmaps, requirements, UML, Design Sprints, backlog and UAT, ROI reviews, Power BI.

Healthcare Informatics

HL7, FHIR, openEHR, SNOMED CT, GDPR; four years of clinical product ownership in a Nordic EHR.

Consulting & QA

KPMG-grade process re-engineering, vendor evaluations, data migration QA, go-live gates.

Industry domains: Healthcare E-commerce & B2B Travel Tech Banking & Finance Government & Utilities
Tools I reach for:

TypeScript · Python · SQL · Next.js · React · tRPC · Payload CMS · Elasticsearch · Apache Airflow · Cloudflare Workers · D1 / R2 / Queues / Durable Objects · PostgreSQL · Drizzle · Redis · Docker · Mastra · E2B Sandboxes · Vercel AI SDK · Better Auth / Auth0 · Vitest · Playwright · Turborepo · GitHub Actions · Figma · Power BI · Jira / Confluence

Bonus for the curious

My brain, as a graph

I keep a personal knowledge base where roles, projects, skills, domains and technologies are one linked graph — the same way I think about systems. Every note is readable, every connection is real.

Drag to explore · scroll to zoom · click any node to read it.

Open the interactive graph →

Contact

Want the full story? Ask.

The graph above covers most of it — but I'm happy to walk through any of it live.

Saskatoon, Saskatchewan, Canada · Open to product & engineering roles across platforms, search and applied AI.