Untangling India's
largest jobs platform

A platform that hadn't moved since 2012

NCS worked, technically. It had features, and if you were determined enough, you could get what you came for.

Every step discouraged you from trying.

1. The mobile app was the desktop site in a phone shell

No adaptation for how people actually use phones. For a platform where a phone is often the only device someone owns, that's not a polish issue.

It's a fundamental failure.

1.1 Legacy home screen and keyword-only job search.

Images

2. Key features were buried or broken

Job fairs, career counselling, application tracking: all present on paper, all hard to find in practice. The information architecture had grown organically for years.

Nobody had stepped back to look at the whole thing.

3. The platform wasn't built for its actual users

NCS isn't for white-collar professionals switching jobs. Its real users are first-time jobseekers, informal-sector workers, people with limited digital literacy, applying from rural areas on basic Android phones.

The design assumed a comfort with digital tools most of them didn't have.

The problem had to be proven before it could be designed for

A teardown, before any solutioning

Before talking to anyone, I tore the existing NCS app apart, section by section: welcome screen, registration, the hamburger menu, search, listings, counsellor discovery. Not a general impression. A factual baseline of what was actually broken.

What public platforms elsewhere had already solved

I benchmarked seven platforms, Indian and international, mapping registration through to job preferences for each. The international public employment services were the more useful reference. They'd already solved for scale and accessibility in ways a profit-driven job board never has to.

Real conversations, not formal interviews

For jobseeker insight, I talked to people around me: friends, family, colleagues, some blue-collar staff at the organisation. Not a representative sample, but honest reactions. The gap between NCS and what people considered normal was consistent every time.

Employer insight came differently: through Ministry references into how job aggregators actually post, and recruiters within my own organisation who gave me operational context no formal interview would have surfaced.

From problems to features

Before touching any design, I ran a How Might We session across both personas: use the Government of India association to build trust? Move past an IA that relies entirely on a hamburger menu? Build in personalisation? Handle a job posting's full lifecycle through to hire?

That became a content inventory, features organised by where they'd live: search, dashboard, preferences, profile. I knew what the product needed to contain, and where, before I designed a single screen.

A few decisions that shaped everything else

A product this large involves hundreds of decisions.

These four mattered most.

1. DigiLocker as the foundation for trust

NCS's core credibility problem: bot accounts, unverified qualifications, employers who couldn't trust what they read on a profile. I made DigiLocker central to onboarding, not optional.

It ties every account to a verified identity from day one, and lets jobseekers pull real education certificates straight into their profile. For a government platform, that's a genuine edge over private alternatives.

3.1 Verified onboarding via DigiLocker.

Loop

3.2 Verified profile data and fetched certificates.

Image

2. Designing for low digital literacy without dumbing it down

The hardest problem on this project was inclusivity: a platform that had to work for someone applying for their first job from a small town on a basic Android phone, and for a professional comparing offers.

Designing for the lowest common denominator frustrates everyone.

The answer was progressive complexity. Every core flow works in as few steps as possible, with sensible defaults and plain-language labels. Multiple CVs, visibility settings, custom cover letters: all discoverable, never in the way.

The AI-powered CV builder came directly from this. Many NCS users have never written a CV. Instead of a blank form, the platform generates one from their profile data using OpenAI, which they can then edit.

3. Search stopped being a keyword and became a match

Job search on the old NCS was a keyword box and a scroll. No intelligence, no link between a jobseeker's profile and what they were shown.

I built matching both ways. Jobseekers see how they stack up against a listing before applying: what matches, what's missing. Employers get candidates ranked by skill fit instead of recency.

Less filtering, more deciding, on both sides.

4. Closing the quality-of-life gap with private platforms

Plenty of users had tried NCS, found it lacking next to Naukri or LinkedIn, and stopped. Not because the jobs weren't there, but because everything around the jobs was worse. No bookmarking. No status tracking. No employer invitations. No visibility controls. No multiple CVs.

I mapped every feature people expected from a job platform into a checklist against what NCS actually had.

The bar wasn't better than before. It was that coming from Naukri shouldn't feel like a step backwards.

Two journeys, shown end to end

The redesign covers 10+ modules across jobseeker and employer journeys. These two flows show the decisions above in context. The remaining modules are summarised further down.

Jobseeker: onboarding to first application

Registration via DigiLocker or mobile OTP, with document auto-fetch cutting setup time significantly. One profile covers skills, work history, preferences, and career timeline. Multiple CVs, each shareable separately, with visibility settings controlling who sees what.

BeforeAfter

‹›

4.2 Reimagined mobile profile.

Image

4.3 Onboarding flow, end to end.

Loop

Career timeline lets jobseekers tell their story so far: education and experience now, with competitions, hackathons, and certifications planned next, to better reflect blue-collar jobseekers too.

4.4 Career milestones, visualised.

Image

4.5 Timeline detail, in motion.

Loop

A personalised home surface based on profile and preferences, with government listings visually flagged by the official emblem. Search by industry, job type, or experience level, with quick filters for full-time, part-time, freelance, or government roles.

4.6 Personalised home surface.

Loop

Improved search with keyword and location inputs, bookmarking, and saved alerts. A "near me" filter surfaces nearby opportunities, especially useful for blue-collar roles, where distance matters more than job title.

Each listing shows a skill match score, so jobseekers apply with full information, not a guess. Applying itself is two steps: confirm contact details, then attach a CV and optional cover letter.

4.9 Two-step application flow.

Loop

Jobseekers can now track application status at every stage, something both formal- and informal-sector users consistently asked for.

4.10 Real-time status tracking.

Loop

4.11 Application detail timeline.

Image

Employer: posting a job to hiring

A guided job posting flow replaces the old single long form, with drafts and templates for repeat hirers and AI assistance for writing job descriptions.

Candidate discovery surfaces profiles ranked by skill match rather than a flat list, with filters for industry, experience, functional role, and education level.

A dedicated job posting management module lets employers track and update each posting easily.

A candidate management section within each posting keeps everything organised as candidates move through stages.

The rest, briefly

One more module shipped alongside this: a standalone CV builder, available to every jobseeker independent of onboarding. It generates a complete CV from profile data using AI, editable section by section, or built fully by hand.

What I couldn't control, and what I chose not to build

Three things made this project harder than a typical redesign.

I'm not the target user.I've never looked for a job on a government portal, filled a form in a language that isn't mine, or applied on an unstable connection. Every time a flow felt simple enough, I had to ask: simple for whom?

I was product owner across design and delivery, not just handing off.Scoping phase one versus phase two was as much a design decision as any screen. Several modules exist in Figma but aren't built yet, held back by delivery scope, not design readiness.

And NCS sits under the Ministry of Labour and Employment. Some integrations were mandated, some workflows had to match existing policy, and anything touching user data cleared a different bar than a private product would.

Within those constraints, here's where the thinking went further than what shipped.

An entirely different starting point for the informal sector

Once the project was confirmed, I kept returning to a harder question: did the redesign still assume more digital comfort than most of NCS's actual users have? Tailors, cap makers, agricultural workers, first-time jobseekers from small towns: these are the people NCS exists for.

So I explored a different approach: Easy Mode. On by default, since it targets the majority of the user base, with an option to switch it off. Language selection upfront. A persistent "Hear" button to listen instead of read. Assisted profile setup. Occupations granular enough to actually reflect who was using the platform.

It didn't move forward. Scope was set, time was limited, and a full rethink wasn't what had been agreed. That was a pragmatic call, not a wrong one. The thinking still carried over: it's why progressive complexity shows up in features like the CV builder.

The matching engine that couldn't be built

Most employer-side candidate discovery is still rule-based filtering: right education, enough experience, compatible location. Binary checks. Achievable twenty years ago, and still what the platform does today.

What I designed instead was a weighted engine scoring candidates across twelve criteria, skill overlap weighted highest at 35%, then experience, location, role alignment, education, and softer signals. Every candidate gets a percentage match. Profiles ranked by relevance, not filtered by eligibility.

"Quick Picks" (surfacing one candidate at a time to shortlist or pass) was the visible layer. The engine underneath was meant to power lists, search, and homepage recommendations across the board.

Neither shipped. The logic couldn't be built within the time and expertise available. This is the clearest gap in the project between what I designed and what got built.

Live, used, and still going

At the time of writing, the core features here are live. The platform keeps evolving, with more modules already scoped for the next phase.

I'm not citing platform-level growth numbers. The redesign is recent, and isolating design impact from everything else happening on a platform this size would be misleading.

What I can say: a platform with a well-earned reputation for being hard to use now has a meaningfully different experience across every core flow, and real people are moving through it today.

Once there's enough data to say something specific, I'll come back and update this.

What this project changed about how I work

I came in as a designer and ended up a product owner. Those aren't the same job.

As a designer, my job was understanding the problem and designing the best answer. As product owner, my job became making sure the right things got built in the right order, by a team that isn't in my head: scoping phase one versus phase two, deciding what gets cut when EPFO doesn't come through in time, deciding when Easy Mode is a future phase and not this one.

That's a different kind of judgment, and this project is where I actually learned it.

The other thing worth saying is how this started. I put together a concept proposal alone and sent it cold to a government ministry, with no brief, no client, no guarantee anything would happen. Six or seven months later, a follow-up arrived, and a few concept screens became a live product used by jobseekers and employers across the country, with me as product owner across the whole thing.