Ajeet Kumar

About

From analytics depth to product ownership

I am a Platform Product Manager with a strong analytics foundation. I started in product analytics, turning behavioral data into decisions for products serving hundreds of millions of users, and now own data and AI products end to end at DeepIntent, a healthcare Ad Tech DSP.

My work sits where data science, engineering and go-to-market meet: forecasting engines, pricing and packaging systems, inventory quality, and experimentation. I am comfortable shipping probabilistic systems, setting guardrails, quality thresholds, fallbacks and clear UX to build trust when outputs are not always 100 percent correct.

I hold an M.Eng from BITS Pilani and am completing the Advanced Management Program in Business Analytics at the Indian School of Business (ISB), Hyderabad.

At a glance

●Now: PM (Data), DeepIntent
●Domains: Ad Tech, Travel, Media, Healthcare
●Stack: BigQuery, SQL, Python, Looker
●Craft: Forecasting, pricing, experimentation
●Education: ISB, BITS Pilani, MSIT Delhi
●Based in: Bengaluru, India

Experience

Career timeline

A decade spanning pharmaceutical analytics consulting, product analytics at consumer-scale platforms, and platform product management in healthcare Ad Tech.

Product Manager (Data)

DeepIntent, New York / Remote Nov 2023 to Present

Ad Tech · Data Science · Forecasting · Deals · ML Models · Marketplace · Inventory Quality · SaaS

Own forecasting, pricing and packaging products across the DSP, driving roadmap with WSJF prioritization on big rocks (HealthFirst Packages, Priority Tiers, Deal Health, CoPilot).
Ship end to end: PRDs, acceptance criteria and success metrics, MVP phasing, LaunchDarkly flag rollouts, Ops-led alpha, Sales demos and PMM-led enablement.
Lead alignment across Engineering, Data Science, Design, PMM, Partnerships and Platform Ops across U.S., EU and India time zones.

Lead Analyst, Product Analytics

Paytm, Noida Sep 2022 to Nov 2023

Travel product (Bus, Train, Flight) · 350M MAU

Owned analytics for three travel verticals; automated 100 percent of a manual strategic-data tracking process on BigQuery for high availability.
Ran 30+ A/B experiments and funnel analysis; owned the Cancellation Protect feature with pricing discovery that lifted flight revenue by 2 percent.
Built a self-serve diagnostic tool at the event-action level for independent root-cause analysis; presented weekly to the Business Head.

Product Data Analyst

Times Internet (Gaana), Noida Jan 2022 to Sep 2022

Music & podcast streaming · 150M+ MAU

Lifted Minutes Per User from 8.6 to 11.2 by unearthing podcast ingestion issues; presented findings to the CEO.
Trained a BQML logistic-regression model on 50M episodes at 91 percent accuracy to detect auto-skipping, stabilizing the consumption KPI.
Stabilized podcast retention by 18.4 percent and lifted monthly auto-renewal by 20 percent through a payment-gateway recommendation.

Technology Analyst

ZS Associates, Gurugram Jan 2017 to Apr 2018

Pharmaceutical analytics · Javelin Pod

Built analytical tools for Fortune 100 pharmaceutical clients and their sales forces.
Worked a U.S. consulting assignment identifying post-merger data management challenges for a global medical device company.

Product Manager (Data) · DeepIntent

Selected product work

Data and AI products shipped across forecasting, pricing, deals and inventory quality on a healthcare Ad Tech DSP.

Forecasting

Chronos Forecasting Engine

Shipped a new forecasting engine with zero downtime and integrated it into the Forecasting tab powering the Patient and HCP Planner. Delivered the SmartBid toggle, inclusion and exclusion lists, HealthFirst packages, input-budget and output-deals features.

Latency: 3-5 min to 8-10 sec Zero downtime launch Removed live ClickHouse reads

Packaging & Deals

HealthFirst Media Packages

Curated open-auction deals across FAST, Spanish, Sports, Audio and Total Health, and built a reusable launch playbook. Ran an internal alpha with Platform Ops, then shipped multi-package support, device controls, and Bid Guidance by creative length with a 12-month Media CPM methodology.

~$800K spend in 8 months ~750 ad groups Strong self-serve adoption

Deals

Deal Health

Designed and shipped the Deal Health UX and ran a pre-release demo for an enterprise prospect (PHM), giving Sales a differentiated story on deal quality and readiness.

$25M annual deal won Enterprise sales enablement

Forecasting UX

Ad Group Forecasting Widget

Designed an in-flow forecasting widget in ad group create and edit that live-updates with CPM, flight dates, frequency cap, geo, device, audience, inventory and creatives. Partnered with Data Science to launch mid-flight views surfacing estimated impressions, spend and incremental opportunity.

Live in-flow estimates Mid-flight opportunity views

Inventory Strategy

Inventory Priority Tiers

Authored the PRD and success criteria for a 3-tier waterfall with pacing, edge-case handling (empty-tier fallback) and org-level save/apply strategies. Finalized UX (tiering bar, tooltips, edit strategy) and kicked off development, including an LLM-assisted guidance experience.

3-tier waterfall LLM-assisted guidance Org-level strategies

Quality & Monetization

Quality, Targeting & SmartBid

Partnered with Data on Jounce signals for premium inventory classification and initiated spend-impact analysis to safely filter low-quality supply. Evaluated SmartBid, an embedding-based pricing engine, quantifying margin uplift and targeting gains, and ran RFP/CRM analyses for DOOH, EHR and Geofencing.

Premium inventory signals Margin-uplift modeling DOOH / EHR / Geofencing

Product Analytics Portfolio

Analytics that moved the needle

Selected analytics work at Paytm and Times Internet: experimentation, funnel diagnostics, ML-driven quality, and automation at consumer scale.

Paytm Travel

Experimentation & Revenue

Ran 30+ A/B experiments across resume-card booking, Price Slasher, text persuasion and user flows. Owned the Cancellation Protect feature with pricing discovery, and reduced homepage DAU drop with an estimated-date-of-travel back-button popup that also fed price-drop notifications.

+2% flight revenue 30+ experiments DAU drop reduced

Paytm Travel

Automation & Diagnostics

Automated 100 percent of a manual strategic-data tracking process on BigQuery, and built an event-action-level diagnostic tool letting Tech and PMs run independent root-cause analysis. Ran fraud analysis and set booking thresholds that protected genuine bookings without hurting business.

100% tracking automated Self-serve RCA tool Fraud thresholds

Times Internet (Gaana)

Consumption KPI Turnaround

Investigated a podcast usage dip, unearthed ingestion issues, and lifted Minutes Per User from 8.6 to 11.2, presented to the CEO. Trained a BQML logistic-regression model on 50M episodes at 91 percent accuracy to detect auto-skipping, then fixed 218 shows to stabilize the consumption KPI.

MPU 8.6 to 11.2 91% model accuracy 218 shows fixed

Times Internet (Gaana)

Retention, Rewards & Search

Stabilized podcast retention by 18.4 percent, fixed a rewards policy to grow multilingual uploads by 6 percent monthly, and recommended PhonePe as priority gateway, lifting auto-renewal by 20 percent. Built a podcast search dashboard and used NLP on customer reviews to auto-tag departments.

Retention +18.4% Auto-renewal +20% Failed search -7%

Research & Publications

Published research & projects

Peer-reviewed and conference work in GPU-accelerated simulation, machine learning and applied computing. View ResearchGate profile.

arXiv 2026

Closing the Social-Semantic Gap: SPSD for Edge-Based Prompt Compression in Cloud LLM Inference

arXiv preprint, 2026

Springer

Simulation and application performance evaluation using GPU, CUDA C and deep learning in TensorFlow

Springer publication

NVIDIA GTCx

Optimizing Simulation using the Graphical Processing Unit through CUDA C

GPU Technology Conference (GTCx) 2016, NVIDIA · Poster presented in Mumbai

IJCA

Analysis of Machine Learning Techniques Used in Malware Classification in a cloud computing environment

International Journal of Computer Applications (IJCA)

Human Activity Recognition via a mobile tri-axial accelerometer using time-domain features and machine learning

Bachelor's thesis research

Blockchain

Ethereum-based smart contracts between patient, provider and payer for insurance settlement

Applied research project

Recognition

Achievements & honors

🎓

UPSC & IFS Civil Services (Mains)

Cleared to the Mains stage of Civil Services (2020, 2021, 2022) and Indian Forest Services (2022), among 1,800 of 1.4M aspirants.

📈

GATE 2016, 92nd percentile

Computer Science, out of over 1 million aspirants.

🏆

Cloud v20 National Research, Top 10

Reached the semi-finals of Unisys' national-level research competition.

📱

Government of India recognition

Letter of Appreciation from NIELIT for a Digital Payments app; departmental scholarship from the Ministry of Home Affairs.

📲

Nokia Millionaire Club

Windows Phone 8 app "Techtree" selected; world ranking of 310 (DVLUP badge). Google-sponsored open-source contributor.

🌟

Published apps, 5-star rated

Android app on Amazon App Store and Play Store; two Windows apps. Third place at a DRDO-sponsored research presentation.

Toolkit

Skills & expertise

Product Management

PRDs & user storiesRoadmapping & WSJF / RICEAlpha / Beta / GALaunchDarkly flagsGTM & enablementOKRs & success metricsStakeholder management0-to-1 delivery

Data & Analytics

SQL & window functionsBigQueryA/B & multivariate testingFunnel & cohort analysisHypothesis testingForecastingLooker & Data StudioTableauMetric guardrails

Ad Tech

DSP / SSP / DMPOpenRTB & PMP / PG dealsPacing & frequency cappingInventory & supply strategyBid shading & modifiersCTV / OLV / DOOHIdentity & consent (UID2, RampID)Brand safety & IVT

AI / ML & Engineering

Python (Pandas, NumPy, scikit)BQMLTensorFlowNLPEmbeddings & pricing modelsRAG & prompt engineeringCUDA C / GPUGit, JIRA, Confluence, Figma