Pratik GuptaHome

Resume

Pratik Gupta

Senior Product Manager · Marketplace Platforms · Catalog Quality · AI-led Automation

Dubai, UAE · +971 52 202 8872 · pratik.gupta46@gmail.com · linkedin.com/in/pratik-gupta-64899531

Summary

Senior Product Manager with 10+ years building and scaling e-commerce marketplace products across the Middle East and India. Currently lead delivery experience and address-quality products at Amazon for 5 country stores in the MEA4 region (Middle East, Africa, Turkey). Hands-on AI/LLM practitioner who personally builds pipelines, prompt-engineering systems, and ML-product integrations end-to-end. Track record of converting ambiguous platform problems into automated systems with measurable revenue: ~$400M annualized gross-sales lift from a single checkout launch, $56M annualized opportunity unlocked, $37.5M gross-sales launch shipped, $835K cost saved from a single India launch, and 10M+ products auto-enriched.

Selected Achievements

  • ~$400M annualized gross-sales lift from optimizing address creation for customers without a saved address.
  • $56M annualized gross-sales opportunity unlocked through an ML-driven product grouping engine across 9 emerging country stores.
  • $37.5M annualized gross-sales uplift from a homepage discovery launch; reframed a $120K ad-revenue objection to unblock the launch.
  • $835K annualized cost savings from Address Creation through Pin On Maps in India (Delivery Success Rate improvement and refund-concession reduction).
  • 10M+ product titles auto-enriched through an AI-assisted content engine (+14 percentage points in title quality, $8M annualized gross sales).
  • 40K+ products regenerated via a self-built LLM pipeline (+1.7% gross sales per session, validated through A/B experiment).

Experience

Amazon · Senior Product Manager, Delivery Experience

Aug 2025 – Present

Dubai, UAE

Lead product strategy for delivery accuracy and address quality across 5 country stores in the MEA4 region (UAE, Saudi Arabia, Egypt, Turkey, South Africa). Partner with engineering, ML science, operations, legal, and external mapping providers (HERE, MapMyIndia, AfriGIS).

Address Creation through Pin On Maps in India · $835K cost savings

  • Led end-to-end launch and experiment strategy from initial release through gap identification, remediation, and successful re-launch. Partnered with MapMyIndia for the mapping service, and built a customer-facing flow offering a choice between map-based and form-based address creation.
  • Diagnosed integration gaps and infrastructure latency issues post-initial launch; drove platform fixes and added user-experience enhancements (rotation, momentum zoom) before the successful re-run.
  • Outcome: 21 basis-point reduction in checkout address-page abandonment overall, 43 basis points for new customers; +8.94 percentage points geocode precision; +0.36% checkout conversion. $835K annualized cost savings from improved Delivery Success Rate and reduced refund concessions; additional $100K from caching infrastructure costs. India address normalization shipped +2.0 percentage points geocode accuracy and +0.8 percentage points first-time delivery success.

Targeted Intervention at Checkout · 8 country stores

  • Designed and shipped a targeted-intervention experience that prompts customers with defective addresses and updates them at checkout.
  • Partnered with worldwide teams to scale across UAE, Saudi, Egypt, Turkey, India, Australia, Brazil, and Mexico — improving delivery reliability without impacting business metrics.
  • Outcome: 52 basis-point reduction in delivery failures; 18% widget click-through rate; 16 basis points reduction in customer-service contacts; 30 basis points reduction in unit refunds; +9 basis points cost-per-order. All guardrail metrics protected (gross sales, checkout conversion, units ordered, Prime signups).

Optimized address creation for customers without a saved address · ~$400M annualized gross sales

  • Launched a guided customer flow that recovers users with no delivery addresses at checkout.
  • Reduced checkout abandonment by 6.21 percentage points in UAE, 6.04 in Saudi Arabia, and 13.48 in Egypt, leading to ~$400M annualized incremental gross sales.
  • Drove 30k+ fewer customer-service contacts and 17.5k+ fewer unit refunds annualized. Aligned engineering, legal, and worldwide expansion teams, scaling globally.

Deliverable-First Address Reranking · Saudi Arabia & UAE

  • Designed an algorithmic improvement for address-card reranking at address page that prioritizes deliverable building-level addresses over street-level proximity.
  • Authored the linear scoring model with deliverable-precision weighting.
  • Outcome: +0.15% Address Creation Success Rate, +0.11% checkout conversion and geocode accuracy lifted from 35% to 42%.

Saudi Arabia National Address integration

  • Drove short-code coverage to 93% through product interventions, navigated regulatory requirements with the Saudi telecom authority, shipped customer-facing notifications, and onboarded third-party logistics publishers — avoiding a potential compliance violation while improving geocode precision for delivery.

Amazon · Senior Product Manager, Middle East - Catalog & Discovery

Feb 2022 – Aug 2025

Dubai, UAE

Owned product and program for catalog quality, AI-driven content automation, product-page evaluation, and homepage discovery across UAE, Saudi Arabia, and Egypt — three of the region's largest e-commerce country stores. Influenced 6 partner engineering teams and a worldwide product organization.

Product Detail Page Evaluator · EC2 + LLM + headless browser

  • Identified the gap where existing siloed evaluators (Title, Attributes, Bullets, Images) operated independently, missing redundancies and cross-element defects.
  • Built a holistic evaluator on EC2 using headless browser for product-page screenshots, S3 storage, and an LLM evaluation engine assessing correctness, redundancy, and cognitive load through 2 distinct shopper personas.
  • Supports multiple product types, 3 languages (English, Arabic, Turkish), multiple country stores, and custom prompt injection. Built this as a PM without a formal engineering background.

ML-driven product variation engine · $56M gross-sales opportunity

  • Took over a low-adoption ML prototype after the previous product owner exited; reframed the problem and authored the V2 product strategy.
  • Designed a multi-tier prediction architecture with confidence scoring; secured engineering allocation and achieved 45% high-confidence prediction rate.
  • Migrated the ML pipeline from spreadsheet outputs to a managed data pipeline — cut processing latency from 8 days to 1 day; lifted partner-team throughput from 600 to 3,000 products per day.
  • Outcome: $56M annualized gross-sales opportunity unlocked. Broken product-grouping share of search reduced to 3.1% (-2.5 percentage points quarter-over-quarter); +4.1% product-page views, +0.46% conversion on remediated products.

LLM-driven product variation validator · $17M projected gross sales

  • Built a prompt-engineering solution to audit the ML grouping engine's low-confidence recommendations using product text and image vision.
  • Achieved 90% precision and 67% recall on the 55% low-confidence bucket through iterative prompt optimization.
  • Integrated end-to-end across 3 partner engineering teams in 3 weeks.
  • Outcome: Met broken product-grouping reduction goal; ~$17M projected gross-sales impact; 5 headcount avoided; lifted high-confidence recommendation rate to ~70%.

Product Summary regeneration · Self-built LLM pipeline

  • Built a complete end-to-end LLM regeneration pipeline as a PM through self-directed learning; published 40K+ regenerated product summaries across 9 product types from prototype to production-scale deployment.
  • Architecture handed to the Science team to scale to 257+ product types, multiplying impact across the org.
  • Outcome: A/B experiment delivered +1.7% gross sales per session.

AI-assisted title quality at scale · 10M+ titles enriched

  • Surfaced low product-title quality (~40% baseline) through a market proof-of-concept; influenced engineering to fund an Arabic-first content quality engine.
  • Co-designed scoring methodology with 5 quality checks; partnered with data engineering on supporting attribute pipelines.
  • Validated a templatized title-generation experiment: +3.5% incremental Ordered Product Sales.
  • Outcome: 10M+ titles enriched across 241 product categories; title quality lifted from ~80% to 93–96%; $8M annualized gross sales.

Homepage discovery launch · $37.5M gross sales · 8 features at zero engineering cost

  • Localized a personalized infinite-scroll discovery experience from another region's tech stack into the Middle East homepage.
  • Re-derived the ad-revenue model from server-side rendering to client-side views, cutting the projected $120K loss objection by 92% and unblocking the launch.
  • Aligned global homepage and ad leadership pre-Ramadan high-volume event.
  • Outcome: +0.45% gross sales = $37.5M annualized. Onboarded 8 additional homepage features in FY24 with no engineering cost = $22M annual gross sales.

Catalog completeness automation (worldwide first)

  • Challenged the legacy 'search-impression-weighted' coverage measure; proposed buyable-product coverage (10x prior scope) as the right discovery metric.
  • Justified cost-to-benefit and shipped a worldwide-first automation.
  • Outcome: 1,600 product types automated end-to-end (99% coverage); double-digit percentage-point YoY coverage gains across UAE, Saudi, and Egypt; 2 country stores reached category-best globally.

Amazon · Program Manager, Strategic Brands (Amazon Fashion India)

Apr 2021 – Jan 2022

Bengaluru

Owned end-to-end P&L, catalog, marketing, and selection for 20+ premium fashion brands.

  • Scaled the strategic brand portfolio to ~1.5% of Amazon Fashion's apparel revenue in the first year.
  • Negotiated 6 brands to commit 2.8% of annual revenue as category co-op marketing for 2021–22.
  • Owned monthly and quarterly brand business reviews; partnered with marketing, supply chain, and category teams on traffic, conversion, and growth experiments.

Amazon · Marketing Manager → Sr. Marketing Manager

Apr 2017 – Mar 2021

Bengaluru

Owned cart promotions, marketing campaigns, and traffic channels for Amazon Fashion India across Jewelry, Handbags, Footwear, and Sportswear.

  • Designed and launched the cashback-based cart promotions program — a first on Amazon India — generating +18% incremental gross sales at 15% efficiency during high-velocity events.
  • Launched Van Heusen Handbags as exclusive Amazon partner — scaled to the #3 handbag brand on the platform within the launch year.
  • Drove paid customer-acquisition and re-engagement for Shoes and Handbags through push, email, and homepage personalization.
  • Built 60+ semi-automated marketing programs and a monitoring dashboard that cut manual effort by ~70%, while co-rolling out gender-based customer targeting with the analytics team across daily merchandising and event campaigns.

Amazon · Category Store Merchandiser

May 2016 – Mar 2017

Bengaluru

  • Executed on-site merchandising for the footwear category.
  • Owned weekly, monthly, and quarterly reporting of page-level performance and drove insights from it.
  • Assisted in on-site execution of several large-scale events and brand campaigns.

Limeroad.com · Product Technical Analyst

Jan 2015 – Apr 2016

Bengaluru

  • Drove on-site product curation and merchandising across trend, material, silhouette, trims, and style dimensions.
  • Assisted in creation and segregation of email target segments.
  • Managed a team of 16 data-entry operators across vendor data collection.

Skills

Product: 0 to 1 product discovery, product strategy, BRD/PRD authoring, roadmap planning, A/B testing, agile delivery, product lifecycle ownership.

AI / ML: LLM workflows, prompt engineering (text + vision), self-built pipelines on EC2 / S3, headless browser automation, hallucination detection, ML product oversight.

Data & Analytics: SQL, ETL, A/B experiment design, financial modeling, root-cause analysis, dashboarding.

Marketing & Growth: Cart promotions, customer acquisition, push/email/homepage merchandising, GTM, brand co-op, conversion optimization, customer segmentation.

Leadership: Cross-functional alignment with director-level stakeholders, partner-team negotiation, programmatic execution at scale, first-principles problem-solving.

Education

Bachelor of Fashion Technology — National Institute of Fashion Technology (NIFT), India · 2010 – 2014 · CGPA 7.93

Download PDF Resume