Product Manager · Data Products · Technical PM

Nivedita M

I'm a Technical/Data Product Manager. At Side I own Sidekick, the revenue platform that is now the standard reporting tool across the company, along with the data pipelines and the AI billing workflow behind it. Before that I ran the roadmap for Dubs, a consumer app with 100k+ users.

LinkedIn ↗ Email me
01 / Who I am

About

Engineering foundation from RV College and AirAsia Tech. Data science depth from a Master's at the University of Colorado Boulder. Product work on both sides: enterprise (B2B) data products for Finance and Ops teams, and a consumer social app.

I've written the code, built the data models and sat in the user interviews myself, so my calls rest on what's actually true. I run discovery before I write specs, and I read the data before I make the call. Most of the hard problems I've worked on came down to finding the real problem before building the wrong solution well.

I recently completed an Executive MBA at IIM Visakhapatnam, graduating Rank 1, and I hold ISB's Professional Certificate in Product Management, CSPO and a Lean Six Sigma Black Belt.

02 / Career

Experience

Side

Full-time · 3 yrs 11 mos

Bengaluru, Karnataka, India · Remote

Technical Product Owner | Data Products

Dec 2025 - Present · 1 yr

  • Cut invoice preparation time by about 60% (45 to 18 minutes per case) for the Central Billing team in the Office of the CFO by documenting the Fee Journal and Timesheet processes and automating them with an AI-assisted billing workflow; about 70% of cases now need no manual work
  • Cut monthly revenue reconciliation from 2 days to under an hour by owning the roadmap and backlog for Sidekick, our revenue reporting platform used by US and EMEA teams across 1,600+ opportunities; it is now the standard reporting tool for the company
  • Designed Sidekick from scratch after spotting a gap in milestone and payment visibility: ran discovery with Finance, Operations, project managers and leadership and wrote the PRDs for one view of project data that surfaces invoice bottlenecks
  • Cut PM-assignment misalignments from ~150 to 10-12 per cycle with a Data Quality Dashboard and a standard data entry template, and saved Finance ~15 days per cycle with a fuzzy-matching pipeline that maps 5,000+ CRM records to D365
  • Restored trust in revenue reporting by turning IFRS 15 rules into product logic with Finance and running a 100+ measure audit, and moved the team to repeatable delivery with Jira planning, versioned releases, QA checks and PRD templates
  • Led the redesign of the platform's data into a bronze/silver/gold structure in Microsoft Fabric with engineering, so new data sources can be added without rework and data issues are flagged instead of silently dropped; I use SQL to query and validate the data at each stage

Junior Technical Product Owner

Jan 2024 - Dec 2025 · 2 yrs

  • Owned product on Dubs, a B2C social app for gamers with 100k+ users on iOS and Android
  • Shipped an AI character chat (Kyle) with guardrails, limited to game topics, reaching a 60% successful conversation rate
  • Launched an AI avatar feature with 43% of generated avatars set as profile pictures
  • Found the strongest retention driver in Mixpanel: users who messaged in their first session retained at 45% vs 20% on Day 15
  • Raised session duration 22% and cut abandonment 15% with a 3-week A/B test of activity-ranked lobbies
  • Grew a sticker pack to 40% of total revenue after replacing an underperforming collection feature
  • Shipped features across media share, voice chat, tagging, stickers and safe-photo workflows, and took competitor-led proposals to leadership

Senior Software Developer

Jan 2023 - Jan 2024 · 1 yr 1 mo

  • Built the Python API layer for Dubs on AWS, within a microservices architecture
  • Tested endpoints in Postman and queried and integrated an existing MongoDB store

AirAsia Tech

Full-time · 2 yrs 1 mo

Bengaluru, Karnataka, India

Software Development Engineer 1

Jan 2021 - Jan 2023 · 2 yrs 1 mo

  • Developed React frontend features and maintained REST APIs for SNAP, a flight and hotel booking product
  • Led OpenCV image compression on upload, at a 5:1 ratio
03 / Case studies

Work Highlights

Enterprise (B2B) data product Data Quality Finance, Ops, C-suite Company-wide standard

Sidekick: Revenue Intelligence Platform

One set of revenue numbers the whole company agrees on.

+

The problem

Finance, Ops and leadership each had their own version of the revenue numbers, and month-end reconciliation took days.

What I found

I sat down with each team. The reports weren't the issue. Every office entered project data differently, so the numbers couldn't agree however the dashboards were built.

What I did

  • Fixed the inputs first. One standard for how projects, quotes, billing and costs get entered, before any reporting work.
  • One data model, a view per team. Finance, Ops and leadership see different screens built on the same numbers.
  • Showed missing data instead of hiding it. Each PM gets a short list of their own records to fix.
  • Made every number traceable. Any figure can be clicked back to its source, so questions get answered on the spot.
  • Automated the CRM-to-ERP matching. About 70% now matches automatically; Finance checks the rest.
  • Moved the sales revenue report onto live CRM data. While checking it against the old report, I found the old one was missing finance bookings the CRM already had.
  • Built in order of hours saved. The features that removed the most manual work went first.

Impact

  • Reconciliation: 2 days → under an hour
  • PM assignment errors: ~150 → 10-12 per cycle; resolution time from ~2 months to ~1 week
  • CRM-to-ERP matching: ~15 days saved per cycle
  • Payroll processing: 2–3 days → about an hour
  • Now the standard reporting tool for the company

Roadmap

Phase 1 - Foundation (shipped)

Phase 2 - AI-integrated insight generation (shipped)

Phase 3 - Seamless end-to-end integration (vision)

B2B AI workflow design Finance automation

Fee Journal Automation

AI reads the billing documents. Rules decide what gets posted.

+

The problem

Finance built the monthly billing journals for the ERP by hand. I mapped both billing routes with Finance, step by step, and asked for each raw export alongside the hand-corrected copy. The difference between the two showed exactly where the manual work was.

Before

1. TriggerOps confirms the work; the rate comes from the signed work order
2. ExportHours or quotes pulled from the source system
3. Fix by handThe same spreadsheet corrections, every period
4. UploadERP prices it and raises the invoice

The design

1

Prepare

Source data lands in Fabric, gets checked, and is frozen for the month.

System
2

Read

AI pulls the billable lines from the approved quote and PDF.

AI
3

Build

Mapping rules create the journal. Validation flags errors.

Rules
4

Review

Finance approves, edits or rejects. Nothing posts without this.

Finance
5

Post

Upload file first, ERP API later.

System
6

Reconcile

Source, journal and invoice must match, or it raises an exception.

Rules

Key decisions

  • AI only reads the documents. Rules do the mapping, so every journal comes out the same way and can be explained in an audit.
  • Finance reviews every journal. The review screen is part of the product.

Impact

  • Preparation time: 45 → 18 minutes per billing case (about 60% faster)
  • About 70% of cases reach a validated upload file with no manual work; the rest go to exception review
  • Went live after a four-week pilot run alongside the manual process with one billing team
Consumer Social 100k+ users Mobile iOS + Android

Dubs

A safer social platform for gamers to find teammates and build communities around the games they play.

+

The problem

Gamers, many of them teens, had nowhere safe to find teammates. Existing apps handled chat once you'd found your people; Dubs set out to help them find those people in the first place.

How I ran product work

How I run product work: Frame, Align, Deliver, Launch and Learn

Core product areas I led

Discovery & Matchmaking

+

What this area was: How users found live lobbies and the right people to play with.

Key decisions I made

  • Reused one ranking model across Hot Lobbies and Autoqueue
  • Kept lobby creation minimal so discovery stayed the default path

Dig into the detail

01
Hot Lobbies
+
Hot Lobbies bannerHot Lobbies list view

Users could see which lobbies were actually active before joining, instead of guessing and bouncing between dead rooms.

02
Autoqueue
+

A guided matchmaking flow that helped users get into the right lobbies faster, without manually scanning a long list of rooms.

03
Team Up
+
Team Up swipe card

A profile-based friend discovery flow designed to help gamers evaluate potential teammates quickly and find stronger matches earlier.

04
Lobby
+
New lobby creation form Lobby home screen

Designed the core live-session experience where users discovered active games, evaluated whether a lobby was worth joining, and coordinated once inside.

AI Experiences

+

What this area was: AI-powered identity and interaction inside a social gaming product.

Key decisions I made

  • Limited AI to two jobs users cared about: identity and conversation
  • Kept the bot in one moderated channel

Dig into the detail

01
AI Avatars
+
AI Avatar Credits generation flow

An AI-powered avatar flow that let users generate a more expressive version of themselves and use it across their profile.

Key decisions I made

  • Put avatars on the profile itself, where users would see and use them
  • Generation runs in the background, so users keep using the app while it processes
  • Credits are charged once per request, and refunded automatically if generation fails

Result: 43% of generated avatars were set as profile pictures.

02
Kyle Bot
+
Kyle bot chat interaction

A dedicated AI character channel that explored how bot interaction could work inside a live social product.

Why this design

Guardrails - Kyle was limited to game-related topics and designed to deflect unsafe or off-topic prompts.

Disambiguation - AI iconography made it clear when users were interacting with Kyle rather than another person.

Result: 60% successful conversation rate.

Communication Layer

+

What this area was: Chat, voice and communities: where users spent their time once they'd found each other.

What I owned

  • Chat - text, tagging, stickers and media in one place
  • Voice - live voice in lobbies and channels, with mute and moderation controls
  • Media share - photo sharing with safety checks
  • Channels - lasting communities, alongside short-lived lobbies

Onboarding & Social Graph

+

What this area was: How a new user became a Dubs user, from age-gated signup and identity setup to building their first social connections.

Dig into the detail

01
Onboarding
+
Onboarding flow

Age-gated onboarding that moved users from signup into the product while handling notification timing, identity setup and referrals.

Why this design

Permission timing - Used a soft prompt before the OS-level notification request because iOS limits the chance to ask again after a decline.

Safety by design - Added guidance during username setup to reduce risky identity choices early.

02
Friending
+
Accept friend request

The social graph layer that connected discovery to retention, helping users turn promising matches into lasting connections across DUBS.

Why this design

Everywhere - friend requests can be sent from search, Team Up, lobbies and chat, because friending was the output metric for Team Up.

High-intent actions first - Prioritized Accept Request in search results to reduce friction on the most valuable social action.

Monetization

+

What this area was: The in-app economy and store.

What I owned

  • Store, the Dubloons virtual currency, and purchase gating for teens

Key decisions I made

  • Economy control - kept purchases tight at launch so the currency wouldn't inflate
  • Teen purchases - parental consent built as a flow teens could complete
  • Store clarity - clear owned and purchased states, to stop accidental repurchases
Store and purchase flow

Moderation

+

What this area was: The systems that made DUBS safer to join, trust, and participate in.

What I owned

  • Age safeguards - teens and adults kept apart
  • Trust Score - visible reputation signals
  • Moderation controls - mute, ban and role permissions in channels and lobbies
  • Reporting and blocking across every social surface

Outcomes

  • Built Mixpanel analytics from 0 to 1 with engineering (event schemas, DAU/WAU/MAU, activation and retention dashboards)
  • Retention driver: users who messaged in their first session retained at 45% vs 20% on Day 15
  • Hot Lobbies A/B test: session duration +22%, abandonment -15%, rolled out to 100% of users
  • Sticker pack: 40% of total revenue
  • AI character chat: 60% successful conversation rate; AI avatars: 43% set as profile pictures
  • User research after every two major releases to feed the next round
Research AI Governance Published

Industry Intelligence Report

Contributing analyst on the XDS 2026 Insights Report.

+

A survey of 250+ studios, publishers and service providers on external development and AI governance. I set up the analysis structure, kept questions comparable year over year, and added new measures on AI permissions, disclosure and data-training limits.

Download the full report ↗

04 / Toolkit

Skills

Product Management

Product strategy · Roadmap ownership · PRD writing · User stories and acceptance criteria · User research · Funnel analysis · Retention strategy · RICE prioritization · A/B testing · Sprint planning · Stakeholder management

Technical Product Management

Technical discovery · Data-flow mapping · Data quality · Semantic modelling · Pipeline design · SQL · Microsoft Fabric · REST APIs · Jira · Git

AI Product Management

AI product design · LLM concepts · AI workflow design · AI safety and governance · Human-in-the-loop systems

Domains

B2B enterprise data products · Consumer social · Finance and Operations · ERP · Trust and Safety

05 / Credentials

Education & Credentials

Education

Certifications

Professional Certificate, Product Management

Indian School of Business (ISB) · 2024

Certified Scrum Product Owner (CSPO)

Scrum Alliance

Lean Six Sigma Black Belt

The Council for Six Sigma Certification (CSSC) · 2024

Awards

3rd Place, NASA International Space Apps Challenge 2020

Local event hosted by SRM Institute (AARUUSH'20) · October 2020

First Prize, Best Project Presentation

R.V. College of Engineering · 2020–21

06 / Get in touch

Let's talk

Open to Technical PM, Data PM, and B2C or B2B PM conversations.

niveditamadhava1@gmail.com · +91 9482127568

Email me Call me LinkedIn ↗

Bangalore, India