Yatin Khar
Open to work · APM/PM, Growth & GTM

Yatin Khar

I build at the intersection of operations, growth, and systems thinking.With 3 years of experience shipping complex 0→1 products in the automotive and defense sectors, I am now translating my systems-first approach into the digital space. I am currently seeking APM/PM and Growth roles to help innovative companies build, scale, and launch.

Delhi, India IST
Open to relocate · Mumbai, Pune, Bengaluru, Remote
Selected Work

Builds & Teardowns

A mix of things I've built and things I've pulled apart over the span of my masters. The builds are 0→1 products taken from concept to working software - designed and developed end-to-end, not yet launched or marketed. The teardowns are structured product analyses - examining user behaviour, business model mechanics, and design decisions to surface prioritised, evidence-backed recommendations.

01
Zepto: Fixing Zepto's Out-of-Stock Checkout Loop
Zepto wins on speed. But the out-of-stock recovery loop costs 5 steps and ~40 seconds - exactly when Blinkit is one swipe away.
2.5M
Orders
Zepto
Zepto
In brief

Zepto's checkout was designed for a world where inventory is reliable. In a dark store stocking 2,500–3,000 SKUs across 2,000 sq ft, it isn't. I audited the three failure points - OOS recovery, Bill Details scannability, and the ₹99 AOV nudge - using Jakob's Law, Gestalt principles, and Nielsen's heuristics, then proposed a targeted fix: the Inline Smart Substitution Toggle.

3 Things I Found
  • 01Each dark store stocks only 2,500–3,000 SKUs - OOS is structural, not operational.
  • 02The OOS recovery flow takes 5 steps and ~40 seconds - exactly when Blinkit is one swipe away.
  • 03₹99 free-delivery threshold creates self-selected AOV lift with no visible subsidy.
Read full teardownRead
Zepto
02
Rapido: How Rapido's SaaS Model Won Driver Loyalty
How switching from commission to a flat driver subscription flipped supply-side loyalty across 500+ cities.
$3B
Valuation
Rapido
Rapido
In brief

Rapido won by flipping the model: zero commission, ₹9–29/day SaaS fee, drivers keep 100% of fares. I analyzed how this supply-side lock-in made them impossible for Ola and Uber to match without destroying their own P&L - and why the $3B valuation follows directly from that asymmetry.

3 Things I Found
  • 01Captain net daily take-home rises ₹200–350 after switching from Uber's 20–25% commission cut.
  • 02Matching-phase abandonment ~23% - the opaque spinner, not supply, is the churn driver.
  • 03500+ cities where Rapido has no real competitor - the $3B bet is on Tier-2, not Delhi corridors.
Read full teardownRead
Rapido
03
Swiggy: How Swiggy Built Its Hyperlocal Super-App
How Instamart, Swiggy One, and Bolt turn a food app into a hyperlocal lock-in engine where each product feeds the next.
₹23k Cr
Revenue
Swiggy
Swiggy
In brief

Swiggy's real bet isn't food delivery - it's owning hyperlocal demand. Instamart, Swiggy One, and Bolt are flywheel plays that compound into a lock-in engine: each product feeds the next, making Swiggy sticky at every wallet size and purchase occasion.

3 Things I Found
  • 01Instamart runs on dark stores - structurally separate logistics from food delivery.
  • 02Swiggy One bundles food + grocery + dining into a single loyalty layer, blocking Zomato at the subscription level.
  • 03Bolt (20-min food) attacks Zomato's core without eating into Instamart's convenience margin.
Read full teardownRead
Swiggy
04
Hikeo: Built a Platform That Plans Your Trek in 40 Seconds
A full-stack AI trekking platform for India - planning, discovery, maps, community, and gear. MVP live. Built end-to-end, solo.
149
Treks
Hikeo
Hikeo
In brief

Hikeo is the single destination for planning a trek in India - from first-timers researching Kedarkantha to experienced trekkers seeking hidden routes in Arunachal Pradesh. It replaces 6–8 scattered sources (outdated blogs, WhatsApp groups, scattered Reddit advice) with one AI-powered platform. Built 0→1, solo, full-stack.

3 Things I Found
  • 01Acclimatization protocol is a hard constraint, not a warning - the AI won't generate an unsafe day-gain.
  • 02Marketplace converts gear intent at the planning stage - highest purchase intent in the trek funnel.
  • 03Discover creates intent; Search serves it - curated trek cards and a comparison panel surface routes users didn't know they wanted.
Read full productReadLive SiteSitePRD
Hikeo
05
KartMetrix: India's Unified Go-Karting Platform
A B2B2C platform to discover and book go-kart slots on web or WhatsApp, track lap stats, and give operators live bookings, kart-health, and revenue tools.
90s
Booking Time
KartMetrix
KartMetrix
In brief

Go-karting in India is a digital island - bookings happen by phone call and walk-in, lap times vanish when the session ends, and operators run on a counter diary. KartMetrix unifies discovery, web + WhatsApp booking, QR check-in, driver profiles, and an operator dashboard with a real-time kart-status grid into one platform.

3 Things I Found
  • 01Book on web or WhatsApp in under 90 seconds, with a QR check-in ticket at the gate.
  • 02Operators get a live kart-status grid, walk-in blocking, and revenue analytics in one view.
  • 03Post-session lap times delivered on WhatsApp with a one-tap 'book again' - the re-engagement loop.
Read full productReadLive SiteSitePRD
KartMetrix
06
LinkedIn: Finding the Best Path to a Warm Intro
A relationship-strength scoring system that surfaces the single best connector for any outreach.
3×
Acceptance
LinkedIn
LinkedIn
In brief

Mutual connections are noise. I designed a relationship-strength scoring algorithm that ranks every connector by recency, interaction depth, and shared context - surfacing the single best path to any intro. Targeting 3× higher acceptance rates vs cold outreach.

3 Things I Found
  • 01Bidirectional DMs in last 6 months weighted highest - one-sided follows are filtered out.
  • 02Tagged-together in posts scores above raw connection count - co-presence beats network proximity.
  • 03Connector 'fatigue' signal prevents routing intros to senior ICs who've stopped responding.
Read full growthRead
LinkedIn
07
CRED: Making High-Value Payments Feel Safer
A labor-illusion processing state that makes high-value transfers feel safer at the moment of transfer.
−₹40L
Support
CRED
CRED
In brief

High-value payments feel risky when they happen instantly. I designed a 2.5-second intentional processing state using the Labor Illusion principle - making ₹25K+ CRED transfers feel safer and cutting an estimated ₹40L in support tickets without touching actual payment speed.

3 Things I Found
  • 01Trigger threshold is ₹25K+ - below that, instant confirmation is the correct UX.
  • 023 animated verification steps shown during the 2.5s - perceived thoroughness, not fake delay.
  • 03Top support ticket category was 'my payment is stuck' - resolved by a confidence signal, not a speed fix.
Read full growthRead
CRED
08
Groww: Reducing SIP Cancellations During Market Dips
A SEBI-compliant behavioral intervention at the cancel screen that reframes fear as opportunity.
−20%
Cancels
Groww
Groww
In brief

January 2025 was the first month in AMFI history where SIP cancellations exceeded new registrations. I designed a SEBI-compliant behavioral intervention at Groww's cancellation screen - surfacing RCA gains and a pause option to reframe panic as opportunity, targeting a 20% reduction in correction-period abandonment.

3 Things I Found
  • 01Screen fires only when Nifty is ≥7% below its 52-week high - targeted, not shown on every cancel.
  • 02RCA gain displayed at cancel: 'your last 3 SIPs bought 18% more units than September.'
  • 03Pause option (1–3 months, auto-resume) fills a product gap Kuvera has had for years - ships independently.
Read full growthRead
Groww
Work
Work