About Projects Thinking Lab Get in touch
Platform & API Products AI/ML Systems Consumer Products · 0→1 Experience · 8 yrs

Shivangi
Tripathi

I build products at the intersection of AI infrastructure and consumer experience, from prompt pipelines and B2B APIs to the user flows that make people stay.

At VoyceMe, I took Fableverse from zero to 150K+ users in ~3 months while shipping the Scene Engine API for external platforms. At Quso.ai, I helped scale a creator SaaS to $2M+ ARR owning roadmap, AI features, monetisation, and lifecycle.

Shivangi Tripathi
150K+
Users scaled
$2M+
ARR contributed
Platform Product Design B2B API Development AI/ML Systems Consumer Products 0→1 Prompt Engineering Infrastructure Model Evaluation Product-Led Growth Activation & Retention Pricing Strategy Cross-functional Leadership Platform Product Design B2B API Development AI/ML Systems Consumer Products 0→1 Prompt Engineering Model Evaluation Product-Led Growth Activation & Retention Pricing Strategy Cross-functional Leadership

AI Platform Products

From training pipelines to prompt intelligence systems to generation APIs, I've owned the full AI infrastructure layer, and the consumer experience built on top of it.

Consumer Products · 0→1

Scaled Fableverse to 150K+ users and Quso.ai to $2M ARR by designing activation loops, behavioural onboarding, and retention mechanics that actually form habits.

Platform & API Design

Built and defined the SEE API, VoyceMe's B2B image generation platform, including request architecture, latency thresholds, client dashboards, and partner onboarding workflows.

Operator Mindset

I think in systems. Whether it's a model evaluation harness or a pricing experiment, I define success upfront, instrument everything, and ship to learn, not to announce.

Who I am

About Shivangi

I'm a Product Manager who's always owned more than the roadmap. In every startup I've worked at, the line between PM and PMM didn't really exist, so I've shipped products, defined their positioning, run their GTM, built their lifecycle, and tracked their retention. All of it.

My edge is holding both ends at once: the architecture and the experience, the API contract and the user motivation, the data signal and the human reason behind it.

Career & Scope

  • PM, AI Platforms & APIs @ VoyceMe Mid 2024–Present. Fableverse 0→150K+ users; Scene Engine ~1,000 images/day; SEE B2B API; internal ML tooling
  • Global Product Manager @ Quso.ai Jun 2023–Jul 2025. $2M+ ARR; AI Writer, AI Carousel, AI Influencer; 25K+ MAU
  • Growth Product Manager @ Tickertape Dec 2021–May 2023. 1.2M+ MAU consumer fintech; iOS launch 41K installs in 3 weeks
  • Growth Product Manager @ DeHaat Oct 2020–Dec 2021. 100K+ rural users
  • Product Marketing Manager @ Classplus Mar 2018–Oct 2021. 4M+ user educator SaaS
View full resume →

My Process

  • Watch before hypothesising Session replays, cohort drop-offs, support patterns before anything else
  • Define success before scoping North star metric, leading indicators, instrumentation plan, all before the PRD
  • Ship the smallest valid test first Zero-eng or one-day proxy before committing to a full feature
  • Own the full problem Discovery, scoping, GTM, adoption tracking, post-launch iteration
  • Retention is designed in, not bolted on Map the full usage loop before scoping anything
  • Fluent in the AI layer Prompt pipelines, model evaluation harnesses, training data workflows

Outside the Product

  • Certified SCUBA diver most at peace 18 metres under
  • Trained Bharatiya Classical singer riyaaz keeps me grounded
  • Digital artist I make art the same way I make products: from scratch
  • Spiritual at the core I feel divinely blessed
  • Fiction & fringe science parallel universes, quantum mechanics
  • Dog mom to two golden retrievers and a husky (Maximus, Loki and Perseus)
  • Chai over everything non-negotiable, no exceptions
My Work

Featured Projects

01 / VoyceMe
Consumer Product

0 to 150K Users by Following What Users Built, Not What We Designed

Fableverse launched with curated story scenarios. Users had other ideas. I followed the behavioral signal, character ownership, message depth, image generation, and rebuilt the product arc around what was actually driving retention.

Consumer Product0→1RetentionUGC
View case study
02 / VoyceMe
B2B Platform

Turning an Internal AI Engine Into a B2B API, and What Partners Broke

The Scene Engine worked perfectly for Fableverse because we knew where all the edges were. I productised it for external chat platforms, and every partner integration revealed an assumption we hadn't known we were making.

B2B PlatformAPI DesignDeveloper Experience
View case study
03 / VoyceMe
AI Infrastructure

Building the AI Infrastructure That Powered Consumer and B2B at the Same Time

Before the Scene Engine could power external partners, it had to become a real platform. I built the quality systems, prompt infrastructure, and training pipeline that turned an internal experiment into something accountable to paying clients.

AI InfrastructurePrompt EngineeringModel Evaluation
View case study
04 / VoyceMe
Consumer Product

How UGC Characters Became Both the Growth Engine and the AI Foundation

UGC wasn't the original plan. Users showed us that ownership drove retention more than anything we'd designed. The creation avatar became the seed image for generation — connecting the consumer product and the AI platform in a way we hadn't anticipated.

UGCRetentionAI Systems
View case study
05 / Quso.ai
AI SaaS

Two User Types, One Broken Onboarding, 69% Conversion Lift

Trial-to-paid was stuck at 14%. Session data revealed two completely different user types being forced through the same flow. I built behaviorally-routed parallel paths, and the high-intent cohort had 83% higher ARPPU than before.

OnboardingBehavioural SegmentationConversion
View case study
06 / Quso.ai
AI SaaS

The Blank Page Problem That Was Killing AI Writer Adoption

Users were leaving the platform for 45–60 minutes to use ChatGPT and Canva. The first AI Writer flopped — blank input caused decision anxiety. A pill-based prompting system fixed the entry point and drove $17.8K MRR within 60 days.

AI FeatureBehavioral DiscoveryMRR Growth
View case study
150K+
Users scaled on
Fableverse
$2M+
ARR contributed
at Quso.ai
69%
Trial-to-paid lift
via segmentation
8yrs
Across AI, fintech,
edtech & creator tools
Education & Credentials

Background & Toolstack

Education

  • PGDM — MarketingApeejay School of Management · 2016–2018
  • M.A. — Tourism & Travel ManagementBanaras Hindu University · 2014–2016
  • BBAUniversity of Lucknow · 2011–2014

Certifications

  • Product Management FoundationAccredian · 2022
  • Design System Bootcamp2024
  • Certified Digital Marketing MasterDigital Vidya · 2020

Languages

  • EnglishFluent
  • HindiNative

Toolstack

AmplitudeMixpanelGA4FullstoryMoEngageJiraFigmaRedashMetabaseBranchUXCamCleverTapN8nSupawallDatabricksTableauGrowthBook

"Your superpower is being able to dive unbelievably deep on a topic and think about all the angles of what can make it successful or a failure."

Dylan, CEO @ VoyceMe

Mental models

How I Actually Think

Data is a signal, not a sentence

I pull funnels, session replays, and cohort drops before forming any hypothesis. But numbers tell you what, user context tells you why. I don't move without both.

The best features feel inevitable in retrospect

If a feature needs a lot of explanation at launch, something went wrong in discovery. I look for behavior that's already happening and remove the friction around it. The best ideas are usually hiding in what users are already doing.

Build the smallest valid test first

Before full scoping, I look for a zero-eng or one-day proxy. If users don't respond to the proxy, they won't respond to the feature either.

Infrastructure is a product decision

Prompt pipelines, model evaluation, API architecture. Every choice at that layer shapes what users experience. I treat those choices like any other product decision.

Monetisation is a UX problem

Paywalls that feel punishing kill retention. The best upgrade moments feel like a natural next step, designed for the moment a user feels value, not the moment it's convenient for the business.

Retention is the real product

Activation metrics are vanity. D1/D7/D30 retention before celebrating any launch. A feature that brings users back is worth 10 features that only acquire them.

Interactive tools

PM Lab

Frameworks I think with. Try them.

ICE Score Calculator

Tool
Impact5
Confidence5
Ease5
5.0
ICE Score — Prioritise if time allows

PM Archetype Quiz

Quiz

When a feature isn't converting, your first instinct is to...

Retention Curve Visualiser

Tool
D1
D7
D14
D30
D1 to D7 drop is 50% — typical for consumer apps. Focus on the aha moment in first session.
Let's connect

Let's talk product

I'm always happy to connect with people building interesting things — whether that's a role, a collaboration, or just a good conversation about what makes something actually work.

shivangit29@gmail.com LinkedIn
Back to Home My Work

All Projects

Case studies across AI platforms, consumer products, B2B APIs, SaaS monetisation, fintech, and growth.

Back

Shivangi Tripathi

Product Manager · Platforms, APIs & Consumer Products · 0→1 Builder · 8+ years

shivangit29@gmail.com
Key Skills
Product Strategy
Product Roadmap
Feature Prioritization
Scope & Trade-offs
Experimentation & A/B Testing
User Flows
AI Product Systems
Product-Led Growth
Pricing Strategy
Platform Product Design
Funnel Analysis
Product Analytics
Cross-functional Leadership
Product Discovery
API Product Development
AI Platform Development
Experience
Product Manager — AI Platforms & APIs
Voyce.me
Mid 2024 – Present

Built the Fableverse AI platform from 0 → 150K+ users, leading 0→1 development of the image generation infrastructure (~1,000 images/day), internal model training and prompt intelligence systems, and APIs enabling external chat platforms to generate context-aware images.

UX, User Experience & Core Flows
  • Led UX definition for homepage discovery and the in-chat user lifecycle, including onboarding questions, narrative pacing, and action affordances (generate, regenerate, edit) that structured the roleplay interaction model
  • Authored detailed UX specifications for modals, edit flows, and generation controls (selection states, constraints, empty/loading states, fallback behavior), ensuring clear design-to-engineering handoffs
  • UX decisions contributed to ~1 hour average app session time and significantly deeper engagement on mobile vs web
Engagement, Retention & Cohort Strategy
  • Shifted roadmap focus to depth-first engagement, identifying message count and image interaction as key intent signals
  • Drove product changes that resulted in 27% of app users sending 50+ messages, 733 avg messages per app user (~30% higher than web)
  • Improved early retention for high-intent cohorts, achieving ~30% D1 retention for 20+ message users and up to 54% D1 retention in smaller cohorts
Data, Event Analytics & Instrumentation
  • Owned product analytics instrumentation in partnership with the data team, defining event taxonomy, schemas, and properties across chat engagement, AI SEE image generation, UGC creation, and activation triggers
  • Designed funnel checkpoints and cohort logic (e.g., 20+ / 50+ messages, image generators, UGC creators) to ensure product questions were measurable and actionable
  • Used event-level insights to diagnose drop-offs, inform prioritization, and guide iteration across engagement and retention loops
AI SEE Image Generation (Core Product System)
  • Owned the product definition and UX of AI SEE image generation as a creator engagement and monetization surface inside chat
  • Defined generation behavior across prompt construction, async delivery, loading states, and fallback UX, balancing latency, model quality limitations, infrastructure cost, and chat continuity
  • Scaled SEE usage to ~1,000 image generations per day with entirely organic traffic, making SEE one of the platform's most-used interactive features
SEE API & Platform Enablement (B2B Extension)
  • Led product definition for SEE API, extending VoyceMe's AI generation platform into a B2B product enabling external chat platforms to integrate real-time image generation
  • Defined API product behavior including generation request structure, prompt rendering from live chat context, async vs real-time delivery, latency thresholds, and error handling
  • Worked with AI engineers to improve model training targets and prompt rendering pipelines, ensuring generation quality across partner use cases
  • Designed client-facing usage dashboards and analytics instrumentation tracking API volume, latency, success rates, and downstream image interactions
UGC Creation Systems
  • Owned UGC character creation as a product system (create → test → regenerate → repeat), driving 63% UGC participation, 44% character creation penetration, and ~3 generations per user
  • Designed and optimized activation triggers inside product flows, achieving 19% sign-up conversion via character creation and 22% conversion on regenerate actions (highest-performing trigger)
  • Identified creator retention constraints (15–18%) and surfaced product-side levers to improve repeat creation and long-term engagement
AI Model Training & Prompt Intelligence Tooling
  • Built internal tooling for dataset curation, prompt intelligence, and generation quality evaluation across the Scene Engine platform
  • Defined workflows for training data ingestion and annotation, including a custom browser plugin for collecting and tagging training images from the web
  • Developed systems for prompt similarity scoring, prompt rewriting, and prompt generalization to improve generation consistency
  • Helped establish model evaluation workflows, including test harness validation and bad-generation classification
Global Product Manager
Quso.ai (formerly vidyo.ai)
Jun 2023 – Jul 2025

Led product roadmap, monetization, and activation initiatives across web and mobile, contributing to $2M+ ARR. Owned feature delivery, lifecycle design, and PLG motion for a 25K+ MAU creator SaaS, combining user research, cohort analytics, and GTM execution.

Product-Led Onboarding & Retention
  • Redesigned onboarding using behavioral segmentation and dynamic routing, improving homepage engagement from ~70% → 92% and first-week activation from 11.5% → 14.1%
  • Introduced guided walkthroughs and coach marks informed by session replay and drop-off analysis, increasing first core action completion by ~28% and feature adoption by 31%
  • Built churn prediction models using RFM and activity scoring across 7- and 30-day cohorts, reducing churn from 17% → 10.2%, and to ~7.8% for targeted cohorts
Feature Development & AI Product Launches
  • Led development and launch of AI Writer, AI Carousel, and AI Influencer; piloted Chrome extension for transcription workflows
  • AI Writer v1 increased free-to-paid conversion by 41% post-launch and added $17.8K MRR within 60 days
  • Shipped Workspace Collaboration features (team invites, approvals), achieving 18.4% adoption among active accounts within 45 days
Monetization, PLG & Pricing
  • Introduced reverse free trial across US & Canada cohorts, improving trial-to-paid conversion from 6.9% → 11.7% (+69%) and adding ~$31K ARR in 90 days
  • Launched feature-based add-ons, improving ARPU by 18% and contributing ~$51.2K in view-through revenue
  • Overhauled pricing architecture with modular, usage-based plans; 78% of paying users upgraded to annual subscriptions
Data, Growth & Platform Scale
  • Designed Amplitude event schema and dashboards, increasing active analytics usage by 28% and accelerating experiment iteration
  • Drove SEO and distribution flywheels (AI tool pages), adding 9.1K users, 2.8K trials, and 42% organic traffic growth over 90 days
Growth Product Manager
Tickertape (Smallcase)
Dec 2021 – May 2023
  • Owned product surfaces across onboarding, discovery, and premium conversion for a 1.2M+ MAU consumer fintech platform, partnering with design and engineering on scope and prioritization
  • Defined and shipped the iOS onboarding and migration flow, enabling 41K installs in 3 weeks and migrating ~79% of active web users within 14 days
  • Owned product definition and rollout of core features (ETF Screener, Signals, SGB Tracker), contributing to ~17% MoM activation lift for newly launched experiences
  • Redesigned onboarding and lifecycle flows using usage-based triggers, improving click-to-conversion from 1.1% → 2.3% across high-intent journeys
  • Informed pricing and upgrade mechanics through cohort and usage analysis, contributing to a 31% improvement in premium upgrade rate
Growth Product Manager
DeHaat
Oct 2020 – Dec 2021
  • Owned onboarding and activation flows for farmer-facing and AI advisory features used by hundreds of thousands of users
  • Drove UX and flow-level changes based on 15+ field visits and usability sessions, improving first-week feature usage by 21% and reducing early churn by 12%
  • Partnered with Tech PMs to define success events, localization requirements, and rollout sequencing, improving feature adoption by ~27%
Product Marketing Manager
Classplus
Mar 2018 – Oct 2021
  • Partnered with product and design to shape onboarding and creation flows for an educator-first SaaS platform scaled to 4M+ users
  • Contributed to product-led activation by improving setup flows and in-product guidance, increasing trial signups by 53% and reducing setup drop-offs by 18%
Education
PGDM — Marketing
Apeejay School of Management, Delhi
2018
M.A. — Tourism & Travel Management
Banaras Hindu University
2016
BBA
Lucknow University
2014
Certifications
  • Design System Bootcamp (2024) — Credential ID: jWpLTWCc
  • Global Certificate in Product Management & Product Management Foundation — Accredian (2022) — Credential ID: INDM16953
Tools
Amplitude Mixpanel GA4 Clarity Jira Branch UXCam Redash Metabase Fullstory MoEngage CleverTap Figma N8n Heap Supawall GrowthBook Tableau Databricks
Back to Home PM Tools

The Lab

Frameworks I think with. All run in the browser, no data saved. Try them.