— WORK / ENGAGEMENT TEMPLATES · NOT A PORTFOLIO

What an Accord engagement can look like.

This page is deliberately not a portfolio. We don't claim past projects here — those live privately on References, shared 1-to-1 under NDA when the fit is right. What follows below are forward-looking engagement templates: common Hong Kong business scenarios, what Accord can take on, the kind of stack we'd reach for, and the deliverables a typical engagement of that shape ends with. Read these as "here's what an Accord project could look like for you", not as "here's what we did".

EVERY ENGAGEMENT STARTS ON PAPER
TEMPLATE · MULTI-YEARF&B · OPS

A multi-outlet F&B chain, re-platformed.

Common scenario:a HK F&B group running their kitchens on a decade-old POS held together by spreadsheets and tribal knowledge. What we'd propose: a fixed-fee discovery to map the ticket-to-table flow, then a short focused build with one to two engineers on-site through cutover. Outcome we aim for: every outlet on one platform, every order flowing through code both teams can read.

Faster
ticket-to-table
One
platform across outlets
Months
to full rollout
Often
retainer follow-on
KotlinSwiftUINodePostgresAWS HK
CASE 01 · F&B PLATFORM CENTRAL CAUSEWAY SHATIN TST KWUN TONG MONG KOK +24 MORE +8 MORE
A network of outlets, on one platform.
TEMPLATE · LOYALTYRETAIL · iOS

A heritage brand's loyalty app.

Scenario: a long-established HK retail brand wants a loyalty app, not a marketing gimmick. What we'd propose: a focused short-cycle build with offline-first stamps, in-store scan, and a member-tier system the founder can edit without ringing us. The point is not the app — it's a loyalty programme the owner controls.

Quality
App Store presence
Owner-edit
member tiers
Higher
repeat-visit rate
Quick
cycle to launch
SwiftUIKotlinSupabaseStripe HK
CASE 02 · RETAIL TIER 01 · BRONZE FREE ENTRY TIER 02 · SILVER 3 VISITS TIER 03 · GOLD 12 VISITS TIER 04 · PLATINUM INVITE ONLY
A loyalty ladder, member-tier UX.
TEMPLATE · PRIVATE RAGFINANCE · AI

A document copilot for a financial advisor.

Scenario: a regulated HK advisory drowning in KYC and policy documents, where sending client data to a third-party LLM is not an option. What we'd propose: a private RAG copilot running on dedicated infrastructure with open-weights models, an evaluation harness to measure quality, and a handover doc so your internal team can run it.

Less
manual review time
HK-only
data residency
Eval-driven
model + prompt choice
Yours
to run after handover
Llama 3pgvectorLangGraphFastAPI
CASE 03 · AI DOCUMENT CHUNK VECTOR ANSWER CITATIONS & PROVENANCE 11,000+ DOCS INDEXED 100% HK-HOSTED
A document pipeline, end to end.
TEMPLATE · EMBEDDED SQUADLOGISTICS · ONGOING SHAPE

An embedded squad inside a HK logistics operator.

Scenario: a HK logistics operator whose ops live in spreadsheets, where the manager knows the bottlenecks but no software vendor has spent enough time on the floor to see them. What we'd propose: a small embedded squad — one or two engineers, same weekly standup, on-site enough days a week to learn the operation by hand. Over months, retire legacy systems and ship the small internal tools the team keeps asking for.

Steady
weekly cadence
Retiring
legacy systems
Long-form
engagement shape
Fewer
dispatch errors
Next.jsPostgresRedisMapbox
CASE 04 · LOGISTICS CLIENT · LOGISTICS OPERATOR OPS DISPATCH FINANCE FLEET ENGINEERING — ACCORDHK — CUSTOMER
Two engineers, embedded inside the org.
TEMPLATE · KDSF&B · ANDROID

A kitchen-floor tablet system.

Scenario: a HK F&B chain needs a kitchen display that survives grease, drops, shift handovers, and the occasional dropped tablet — software that has to keep running while staff turn over. What we'd propose: a deliberately simple Kotlin / Room / WebSockets stack, designed for offline resilience first, polish second; deployed via the chain's own MDM so devices replace cleanly without us in the loop.

At-scale
device rollout
Long-life
in production
Fewer
mis-ordered tickets
Stable
no major outages
KotlinRoomWebSockets
CASE 05 · F&B KITCHEN TABLET TICKET 47 · T-12 TICKET 48 · T-04 TICKET 49 · T-09 TICKET 50 · T-07 2:41 0:48 0:12 NEW + ORDERS
A kitchen screen, ticket flow.
TEMPLATE · DISCOVERYLAW · AI

A discovery tool for a HK law firm.

Scenario: a HK firm with multi-language email archives where junior associates burn weeks on early-case review. What we'd propose: a privately-hosted retrieval system combining BM25 with dense vectors, EN / 繁中 / 简中 tokenisers tuned for HK content, and citation-grounded answers so reviewers can always trace back to source. Hosted entirely on the firm's own infrastructure.

Faster
early-case review
Tri-lingual
EN · 繁 · 简
Cited
answers, traceable to source
Private
firm-hosted infrastructure
Qwen 2PostgresBM25 + denseHong Kong dedicated cloud
CASE 06 · LAW CASE ROOT EMAILS CONTRACTS FILINGS MATCH FOUND 240K+ EMAILS · 3 LANGS · HK-HOSTED
A tree of evidence, searched.

Six engagement templates above. For an actual reference matched to your sector — under NDA, with the client's permission — write to us and we'll point you to the quiet references page directly.

/ / YOUR PROJECT HERE

Have a brief worth scoping?

We're booking projects for Q3 2026 onward. Tell us what you're trying to ship and we'll tell you whether the timing makes sense.

Start a conversation → See how we work
✆