Design has a responsibility to the people it claims to serve. Pakistan's Khwaja Sira community has been excluded from every digital platform that exists. Noor is an attempt to change that: a platform built from 16 research sessions, in Urdu and Punjabi, with the community itself.
نور — Light
Overview
Pakistan's Khwaja Sira — the Third Gender, legally recognised by the Supreme Court since 2009 — has been building its own kinship structures, spiritual traditions and informal economy for centuries. What it has never had is a piece of software designed for it, by people who actually spoke with it.
One attempt exists. DhanakApp launched in 2024 as a generic "LGBTQ+ and minority" app, built without community research. It has ten downloads. That failure is not a footnote in this case study — it's the starting brief. A platform built on assumption, however well-intentioned, is indistinguishable from one that was never built at all.
Noor started from the opposite instinct: spend the first four weeks not designing anything, and instead find out what a trustworthy product would even need to do.
Scope · Read this first
This distinction shaped every decision that followed, so it comes before the process. It is the single most important framing choice in the entire project.
Khwaja Sira identity is a distinct, pre-colonial South Asian cultural and spiritual identity — not the Western LGBTQ+ construct. Their history dates to the Mughal era, when they held respected court positions and spiritual roles at Sufi shrines. They have been legally recognised as a Third Gender in Pakistan since the Supreme Court ruling of 2009 — years before most Western trans-rights legislation existed. Grouping them under a Western LGBTQ+ umbrella erases their specific cultural identity, their guru-chela kinship system, their Sufi spiritual traditions and their own internal language, Farsi Kalam. DhanakApp tried the generic framing. It reached ten downloads. Noor is built from sixteen direct research sessions with the community itself.
Khwaja Sira خواجہ سرا — Mughal-era court term, the most respected in Pakistan today
Hijra ہجڑہ — South Asian umbrella term, used across Pakistan, India and Bangladesh
Khusra خسرہ — Punjabi term, informal register
Zenana — effeminate male, a distinct sub-identity
Mukhannas — an Islamic theological category
Guru-chela kinship — a structured dera household led by a guru
Farsi Kalam — the community's own internal language
Sufi spiritual tradition — shrines, blessings, spiritual authority
Pre-colonial recognition — respected advisors in Mughal courts
Legal status since 2009 — a Supreme Court ruling
80% denied formal employment (UNDP, 2024)
92% face healthcare discrimination (FDI & Blue Veins, 2018)
56% depression prevalence (PJMS, 2022)
0 dedicated digital platforms that actually work
10 downloads — the only prior attempt
Before the research
Noor didn't start with a feature list. It started with things I'd never examined closely enough.
Growing up, a Khwaja Sira person visited my village at weddings, festivals and births — welcomed into homes, known by name, part of the community. Later, in bigger cities, I saw a different kind of visibility: begging at traffic signals, performing at functions, working in conditions that looked unsafe or humiliating. I didn't yet understand the systems behind what I was seeing — I just wondered why people didn't find different work, or challenge how they were treated.
For nearly a year I rode Lahore's Speedo buses, and regularly watched Khwaja Sira passengers get stared at and mocked for doing something as ordinary as commuting. One day I intervened, and ended up talking with the passenger in Punjabi and Urdu until their stop came. The questions I asked were shaped by my own limited understanding — I thought they were practical questions. I later realised they placed the burden on the person experiencing exclusion, not on the systems producing it.
That conversation wasn't research — one person can't represent a community. But it gave me a reason to investigate my own assumptions before trying to design an answer for anyone else's life.
Discover · Secondary research
Before I asked anyone to trust me with their story, I did the reading. Twenty-four secondary sources — government census data, peer-reviewed clinical studies, NGO field reports, and 2024–2026 legal analyses — to arrive at interviews already understanding the landscape, instead of asking participants to educate me from zero.
The numbers alone explain why a platform like this cannot be a side project. This is a population living with compounding, documented exclusion across employment, healthcare, education and legal recognition.
Primary journalism, peer-reviewed research, and institutional reports that shaped Noor's design decisions.
Pakistan passed the Transgender Persons (Protection of Rights) Act in 2018. The bigger design lesson from reading it wasn't the law itself — it was that having a right and being able to exercise it are two different experiences.
Ideate · Before primary research
My earliest hypothesis was a private platform where the community could learn, earn and connect without hiding. Before designing it, I had to challenge whether it should exist at all.
Facebook, Instagram, WhatsApp and Reddit already had reach, familiarity and mature tooling. A separate platform risked isolating the community further instead of including it. The gap wasn't that mainstream platforms had no safety features — it's that safety and local context were secondary settings, not foundations.
Discover · Primary research
I ran semi-structured interviews with 16 Khwaja Sira community members across Lahore, Karachi and Rawalpindi, recruited through guru networks and NGO partners rather than open calls — trust travels through existing relationships in this community, not through recruitment forms.
Every session was conducted in Urdu or Punjabi. None were recorded. Handwritten notes only, transcribed and anonymised immediately afterward — a methodology decision, not a convenience. Participants had documented reasons to distrust anything that could be screenshotted, leaked or used against them, and a recording device in the room would have undermined the research before it started.
Refined after the first two sessions, once it was clear a few questions read as interrogation rather than conversation. Each session moved from safe, everyday ground toward more sensitive territory — never the reverse — and no participant was asked to explain or justify their identity at any point.
Taking part in research doesn't automatically grant permission to publish someone's identity, face or personal history. Every quote in this case study is anonymised and reconstructed from handwritten notes, not a verbatim transcript — and any recognisable imagery would need separate, explicit consent for portfolio publication. The visual story should show artists, learners, workers and community leaders — not a sequence of portraits designed to prove hardship.
Before the interviews, I'd noticed some participants behaved differently across social situations, and my early read was that this was simply identity. Research suggested something else: behaviour can be protection. What looks confrontational to an outsider can discourage harassment. What looks like withdrawal can be a way of staying safe.
Discover · Key research findings
These weren't the findings I expected going in. Several of them — the trust deficit above all — inverted assumptions I'd carried from every other consumer product I'd designed before this one.
Every single participant expressed deep suspicion of digital platforms before engaging, grounded in documented algorithmic hostility and personal experience of doxxing and screenshot-based harassment.
Tailoring, embroidery, beauty, digital art, cooking — the skills exist. Identity-based exploitation at the point of transaction blocks every existing marketplace for this community.
CNIC rights, employment protections, education rights — mostly unknown to the majority, corroborating existing Williams Institute research showing very low community awareness of the Act.
Most of Pakistan's Third Gender population lives in urban areas. Migration is common; the dera social safety net does not travel. A 2025 peer-reviewed study confirms that relocating Khwaja Sira lose their entire support structure and restart from zero.
All sixteen participants described depression, anxiety, or trauma. All described avoiding professional mental health support after hostile experiences — consistent with the 92% healthcare discrimination rate documented in secondary research.
Six of sixteen had no education beyond primary level; four never attended school. Independent research documents over 90% of trans students in Rawalpindi/Islamabad reporting institutional discrimination. Physical classroom attendance is not a viable design assumption.
The most useful output of research wasn't a features list. It was a change in how I framed the problem itself.
| Initial assumption | What research revealed | Effect on Noor |
|---|---|---|
| Why don't people defend themselves? | Speaking up can carry real social, economic and physical cost | Safety and control became product requirements, not expectations placed on the user |
| Why don't people know or use their rights? | Information can be legally available while staying fragmented, outdated or unsafe to act on | Madad focused on plain language, sourcing and qualified referral |
| Why do people depend on informal work? | Exclusion from education, employment and finance restricts the choices available | Hunar focused on existing skill and safer market access, not new training |
| Do visible behaviours reflect identity? | Assertiveness, withdrawal and adaptation can also be survival strategies | The product stopped treating behaviour as something to correct |
Define · Synthesis
Affinity mapping turned scattered notes into a shape I could design against. Six clusters emerged, ranked here by volume of evidence rather than by any assumption I walked in with.
Define · Problem statement
Everything after this point in the process traces back to this statement. Every feature that made the MVP had to answer to it directly.
Pakistan's Khwaja Sira / Hijra / Third Gender community have real skills, deep community bonds, and a desire for dignified participation — but lack a trusted, safe digital space to learn, earn, connect, and access their rights, without risking their safety, privacy, or economic precarity.
Define · Decision-making principles
Everything from the IA to the visual system had to answer to these — not just the problem statement.
Define · Personas
I resisted collapsing sixteen participants into a single archetype. The range of need was too wide — a 17-year-old wanting to finish school and a 38-year-old guru managing a twelve-person household need genuinely different things from the same product.
Ideate · Competitive analysis
Four categories of "existing solution" turned up in research: a failed app, a violence-reporting NGO tool, an Indian platform with no Pakistan context, and WhatsApp groups that infiltrate and dissolve. None of them cover the ground this community actually needs covered.
| Platform | Education | Marketplace | Community | Safety | 3rd Gender specific | Gap |
|---|---|---|---|---|---|---|
| DhanakApp (BizzInnova, 2024) | ✗ | ✗ | △ | ✗ | ✗ | 10 downloads · LGBTQ+ umbrella · zero community research |
| Trans-Muhafiz (Blue Veins) | ✗ | ✗ | ✗ | △ | △ | Violence reporting only · KPK focus · no life features |
| PeriFerry (India) | △ | △ | ✗ | ✗ | ✗ | Not mobile-first · no Pakistan context |
| WhatsApp groups | ✗ | ✗ | △ | ✗ | △ | Ephemeral · infiltrated · no structure · no safety |
| Noor ← this product | ✓ | ✓ | ✓ | ✓ | ✓ | First mover. Research-backed. DhanakApp proved generic fails. |
Ideate · Information architecture
The IA is not a features list dressed up as a sitemap. Each of the five tabs maps directly onto one or more research findings — there is no module in Noor that exists because it's a common app pattern rather than because a participant needed it.
A community-specific space creates a real tension: open access lets hostile outsiders in; mandatory legal ID excludes people with incomplete records, or good reason to avoid linking government identity to a sensitive platform. Access strengthens only as the risk of the action increases.
Ideate · Key user flows
Onboarding, the marketplace transaction, and a feature that isn't really a feature at all — a safety toolkit that makes every other screen feel safe enough to open.
Language select (Urdu shown first) → privacy promise, three screens before any data is collected → display name, not required to match legal identity → visibility setting, shown with visual examples → home feed, no further gates.
Key decision — privacy controls appear before any data is asked for. This inverts the standard onboarding pattern deliberately.
Create listing with a suggested auto-watermark → set a fixed price or accept offers → an inquiry arrives showing the buyer the work, not the seller's identity → accept and escrow initiates, holding payment → deliver, and payment releases after a 7-day moderator-reviewed window.
Key decision — escrow is mandatory, not optional. It prevents the non-payment exploitation documented directly in research.
A persistent shield icon stays visible on every screen → one tap hides the profile and all activity instantly, no confirmation step → a pre-configured trusted contact is alerted with location → a crisis resource displays immediately, no navigation required. An early version explored an OS-level power-button trigger; technical review showed a third-party app can't reliably guarantee that across Android devices, so the interaction was rebuilt inside what the product can actually promise.
Key decision — this isn't a feature. It's the feature that makes every other feature feel safe enough to use at all — and it should never claim more safety than it can deliver.
Ideate · Prioritization
Sixty-four candidate features came out of ideation. I prioritized by asking one question of each: which of the six research findings does this address, and does the product still work without it? Anything that couldn't answer that question honestly moved to "Won't have."
Early concepts were useful because they exposed assumptions — some didn't survive contact with research or technical reality.
| Early direction | Why it looked useful | What made it weak or unsafe | Decision |
|---|---|---|---|
| Open map of dera locations | Easy local discovery | Could expose sensitive locations and the people connected to them | Rejected — controlled introductions instead |
| Mandatory legal identity at entry | Discourages hostile outsiders | Excludes people with incomplete documents; increases fear | Rejected — layered access instead |
| Launching every module at once | A complete concept on day one | Requires running a social network, legal hub, payments and a school simultaneously | Reframed — four phases |
| OS-level power-button panic trigger | Fast, discreet, works with the screen off | A third-party app can't reliably guarantee that behaviour across Android devices | Reframed — persistent in-app safety toolkit |
| Direct Urdu translation of "escrow" | Kept the correct terminology | No natural Urdu equivalent, and financial jargon risks reading as a scam term | Avoided — visual step-by-step explanation planned instead |
Design · Visual system
Colour was a research decision before it was an aesthetic one. Every hue traces back to Mughal-era South Asian textile tradition — the community's own visual heritage — rather than any imposed external symbolism. Typography carried the same weight: Nastaliq over Naskh, even though it's technically harder to implement, because it's the script this community actually reads.
Nastaliq is the culturally authentic script traditionally used for Urdu — newspapers, signage, literature. Choosing Naskh, the technically simpler option, would have signalled developer convenience over community respect. The overhead is deliberate.
DM Sans, weights 300–600. Body copy is enforced at a 16sp minimum — a deliberate accessibility floor, since several participants used low-cost Android devices with worn, hard-to-read screens.
Design · Wireframes
Low-fidelity wireframes were sketched in Urdu before any pixel was pushed, referencing the budget Android hardware participants described using in interviews — not a design team's flagship phones. Four screens carried the highest research weight. These are design decisions grounded in what participants said, not results from testing the wireframes themselves — that step hasn't happened yet.
Test · Validation plan
Everything through prototyping was built from and checked against the 16 primary interviews and the affinity-mapped findings. What hasn't happened yet is formal usability testing with participants — no moderated sessions, no measured SUS score, no task-completion data. I want to be direct about that rather than let a testing section imply otherwise.
The reason is the same one that shaped the interviews in the first place: testing this responsibly — in Urdu, with the community, on their actual hardware, without creating new exposure risk — needs the kind of ongoing trust and safeguarding a solo case study can't provide on its own. That's part of what a development partner is for.
External context · Journalism and documentary coverage
Noor was not designed in isolation from current events. The research period coincided with documented attacks, legal challenges to the 2018 Act, and ongoing international journalism covering Pakistan's Khwaja Sira community. These are not Noor's sources — they are the reality any responsible product must acknowledge.
References
Scope · What's real, and what's still needed
Every promise on this screen depends on work behind it — this is where service design stops being a mockup skill and becomes an operating question. Noor touches sensitive identity, legal, healthcare, community, financial and safety contexts. It can't launch as an ordinary independent app.
| User-facing service | Work behind the screen | Responsible partner |
|---|---|---|
| Legal rights article | Source review, plain-language editing, continuous updates | Qualified Pakistani legal partner |
| Healthcare directory | Provider verification, correction process | Healthcare or NGO partner |
| Peer navigation | Listener selection, training, supervision, safeguarding | NGO and mental-health specialists |
| Community circles | Membership rules, moderation, incident response | Community governance team |
| Marketplace | Seller checks, fraud and dispute operations | Community governance + regulated financial partner |
| Learning pathway | Curriculum, assessment, recognised certification | Government or authorised education partner |
| Identity verification | Consent, minimum-data exchange, audit | Approved institutional pathway |
Noor isn't an argument that the Khwaja Sira community should stay separate from wider society, and it isn't built on a belief that the community can't use mainstream technology. It isn't meant to correct anyone's behaviour, rescue anyone, or speak on their behalf. It's a claim about foundations: what digital participation could look like if local language, privacy and ambition were the starting architecture instead of edge cases — not a claim that an app can fix structural discrimination.
Reflection · What changed in my practice
Noor is complete as a design artefact: research, information architecture, personas, a documented design system, and dev-ready handoff notes. It has not been usability-tested, and it is not yet a shipped product — I'm not going to dress either of those up as anything other than what they are.
Three habits from this project changed how I approach every subsequent engagement: recruit through trust networks, not open calls; treat "not recording the session" as a design decision, not a limitation; and let research findings veto features, not just inspire them. The trust-deficit finding alone inverted an onboarding pattern I'd used unquestioned on a dozen prior products.
Given the chance to continue this work, the priority is finding the right partner — an NGO with existing community trust, a government programme like Punjab's Pehchan initiative, or grant funding sized for an 18-month build — because a platform this sensitive should not be built and operated without accountability to the community it serves.
Not more polished screens — the partnerships and field evidence needed to decide whether this can become a responsible service: a compensated community advisory council, participatory threat modelling and an independent privacy review, an unmoderated field pilot of Phase 1, and a partner willing to own legal, healthcare and moderation work for the long term. Success here means safer access to meaningful choices — not download counts.
Noor is seeking NGO, government, and grant partners to develop and operate it responsibly. If that's you — or if you have a different problem worth this kind of process — I'd like to hear about it.
imaasimsharif@gmail.com