Divyansh. Get in touch
Open to product roles · available now

I own the product and the numbers under it.

Growth at Flipkart, data platforms at Meesho, and two products I started myself. I've owned roadmaps and the instrumentation under them — and called time on two builds that weren't going to scale.

Scroll
Divyansh Sharma
Hello
IIT Roorkee '23
B.Tech, Metallurgical & Materials Engineering
Flipkart → Meesho
Ten growth levers, then the tables every team builds on
2 ventures founded
Causal and SummonLM, both co-founded and shipped
10 case decks
Competition wins that turned into Accenture & EY internships

I'd rather ship the smaller thing this month and know whether it worked than ship the bigger one next quarter and guess.

I studied Metallurgical & Materials Engineering at IIT Roorkee and went straight into analytics at Flipkart, where the interesting part was never the dashboard — it was the argument the dashboard settled.

That pulled me into product. Two years at Flipkart, then Data Platforms at Meesho, then nine months building two products of my own to find out what I'd do without a company around me.

See what I've worked on →
Work experience

Four places, two of which I started.

Marketplace growth, internal data platforms, and two products taken from an empty repo to launched.

Data quality checks tab in Meesho's Metric Store, with a failed null check alert
MeeshoApr — Oct 2025
Associate Product Manager II · Data Platforms

Cut detection of a broken core table from two days to same day. Built the function from scratch, 5 tables to 16.

Platform PMData modellingInternal tools
Read about my work→
FlipkartJun 2023 — Apr 2025
Assistant Manager · Business Analyst, Growth

Took exchange adoption from ~12% to ~15%, after thirteen months getting a platform team that didn't report to me to instrument the funnel that showed why.

GrowthExperimentationSQL & BI
Read about my work→
 Wound down
Causal: the Safe Buckets vault marketplace behind a market page showing the price chart and the causal graph of factors driving a box-office market
CausalFeb — Jun 2026
Co-founder · Prediction market intelligence

Shipped a fund product to 30 users — six strategies, four of them profitable. The product worked; the market didn't.

Zero to oneFintechPost-mortem
Read what happened→
 5 pilots
SummonLM analytics: share of voice across AI answer engines, perception over time, mention rate and top citations
SummonLMOct 2025 — Jan 2026
Co-founder · AI-search visibility analytics

Turned twenty leads into five pilots, and found buyers wanted the technical checks, not the dashboards we pitched. Called it when it wasn't going to venture scale.

Zero to oneB2B SaaSGEO
Read about this build→

Both my internships — Accenture Strategy and EY — came out of winning their case competitions rather than a CV screen.

Accenture Strategy · 2022 EY · 2022 Explorin, founder's office · 2022
Product thinking

Case studies

Aha moments in a sales demo — PM School — case study deck, cover page
PM SchoolSmartCue

Aha moments in a sales demo
Where Jindal should expand — Indian Case Challenge — case study deck, cover page
Indian Case ChallengeJindal Stainless

Where Jindal should expand
Three markets, three strategies — EY CAFTA Championship — case study deck, cover page
EY CAFTA Championship2021

Three markets, three strategies
A BUY call on Grasim — R-Street Investment Challenge — case study deck, cover page
R-Street Investment Challenge2021

A BUY call on Grasim
Venezuela's energy revival — VAPA — case study deck, cover page
VAPAVenezuela

Venezuela's energy revival
A fragmented kitchen market — Indian Case Challenge — case study deck, cover page
Indian Case Challengearttd'inox

A fragmented kitchen market
See all 10 case studies  →
What I work in
SQL
Metabase
Mixpanel
Power BI
Experiment design
JIRA
Figma
Claude Code
Next.js
Python
FastAPI
PostgreSQL
How I work

Research is what buys the speed.

01

Research until the answer is obvious

Consumer, market and experts — all three, before anything gets built.

SummonLM started with a co-founder watching a real company's impressions halve in a single core-update week. Two weeks validating that before a line of code.
02

Cut it to the smallest thing that settles the question

The scope that proves the bet fastest, not the scope that covers everything.

At Meesho I scoped a three-phase data quality program and shipped phase one. Phases two and three were the right ideas; phase one was the one that would prove it worked.
03

Then move fast, because the thinking is done

Execution is quick when the arguments happened up front instead of mid‑build.

At Flipkart the exchange fix was a session‑state change. Thirteen months of instrumentation to be able to see it, and simple once we could.
04

Read the result honestly, including when it says stop

The number either moved or it didn't, and I would rather say which.

Both ventures ended on purpose. SummonLM never tracked whether pilot rankings moved — so I can't claim the product caused anything.
Strengths

Five things I'm usually brought in for.

Zero to one, before the spec exists
Two of the four things above started as an empty repo and a hypothesis. I'm comfortable in the stretch where nobody can tell you what to build, and comfortable killing it when the evidence says to.

Experiment design, read honestly
Two years of growth work at Flipkart taught me that most A/B tests fail at the design stage, not the analysis stage. I care about what a test can actually prove before anyone runs it.

Data platforms and the plumbing under the product
At Meesho I owned modelling, lineage and internal tooling. Unglamorous surface area, but every roadmap downstream of it is only as good as whether people trust the numbers.

Structuring a problem fast, under time pressure
Ten case competitions is a lot of reps at taking an unfamiliar industry, sizing it from public data, and landing on a defensible recommendation inside a weekend. Two of them turned into internships.

Building it myself, not just specifying it
Two things I took end to end: the technical checks inside SummonLM and the roadmap they fed, and Safe Buckets at Causal. I owned design and the frontend at both. I build with agentic coding tools now, including this site, so I can prototype an idea instead of describing it — and I know where those tools stop being useful.See the SummonLM build →See the Causal build →
Open to product roles · available now

Want someone who runs it like they own it?

divyansh261001@gmail.com
Divyansh Sharma

30 min call

  • 30 min
  • Cal Video
  • Your time zone

Selected day
See open times

Times open on Cal.com, shown in your time zone.