Three years of turning ₹1L a month into ₹65L, rebuilding a marketing engine mid flight, and shipping my own tools when engineering had a longer queue than my patience. Currently doing this at Classplus. Open to freelance projects and full time roles where the problem is genuinely hard.
I don't see performance marketing as running ads. I see it as building the machine that makes the ads worth running: targeting logic, landing pages, attribution, and the follow up nobody remembers to automate.
I've built that machine from nothing at Bosscoder, and rebuilt one that had quietly fallen apart at Classplus. Somewhere between marketing and growth is where I actually live. Comfortable in an ad account. Equally comfortable shipping my own tools when engineering has better things to do.
Outside the day job, I tear down products I have no professional reason to touch. Purely because the puzzle is fun.
Two companies, one steady climb from figuring things out to owning the P&L.
Find me on LinkedIn
Five things I'm hands on with. Not a menu of everything that could loosely be called marketing.
Google, Meta, and YouTube campaigns built around what a customer is actually looking for, not just where the budget goes.
Query-matched pages instead of homepage traffic dumps, tested against real behavior data, not guesses.
Full-funnel tracking that tells you which channel actually closes deals, not just which one fills a leads report.
Messaging built from what a sales team actually hears on calls, not what sounds good in a strategy deck.
Internal tools and workflows shipped fast with AI-assisted development, without waiting on an engineering queue.
Fair. Most projects start with a diagnosis, not a fixed scope. That part's free to talk through.
One had nothing built yet. The other had everything built, just quietly on fire. Pick your favorite kind of chaos.
Bosscoder AcademyTwo campaigns, side by side. "Full stack development course" pulled cheap, high volume leads from college students. "System design course" cost more per lead but converted working professionals at a much higher rate. That gap changed how I thought about every targeting decision after it. Cost per lead alone was telling me almost nothing.
I introduced campaign and behavior specific landing pages instead of one homepage for everything, studied real user behavior with Microsoft Clarity, and built full-funnel attribution (GTM, GA4, Clarity, LeadSquared) so campaigns could finally be optimized against closed deals, not just form fills.
At ₹1L/month I was running a handful of campaigns and doing most of the legwork myself. By ₹65L/month I'd decentralized execution, picked up two direct reports, and was spending more time on funnel quality than on the campaigns themselves.
Once the ad engine was humming, the ceiling stopped being "more budget" and started being "the signup flow itself." I sat with the tech team directly, mapped where users actually dropped off between clicking an ad and completing enrollment, and pushed for changes that had nothing to do with marketing on paper: form field order, page load speed, the exact moment a counsellor's phone number appeared. None of that shows up in an ads dashboard, but all of it shows up in ROAS.
Every new page went through the same loop: ship a hypothesis, watch real sessions in Microsoft Clarity, not just the aggregate numbers, then decide whether the drop-off was a copy problem, a trust problem, or a plain confusion problem. Most "conversion rate" advice treats every drop-off the same. It isn't. A user leaving because they don't believe the price is different from a user leaving because they can't find the enroll button, and the fix for each is nothing alike.
Build the attribution layer in month two, not month ten. I spent longer than I should have optimizing toward leads before I could see which leads actually turned into revenue. Once that was fixed, everything downstream got sharper faster.
ClassplusBefore building anything new, I audited active spend line by line and found campaigns left running long after they'd stopped performing, plus money going to tools nobody was using. I paused what wasn't working and rebuilt the channel mix around it, leaning harder into lookalike audiences and testing faster.
Biweekly sales feedback sessions surfaced that prospects weren't hesitating on price, they were hesitating on effort and risk. The old line was "Get your own app in ₹19K." The new one: "Launch your coaching brand in less than 10 minutes." Same product, different promise. AOV grew 26% off reduced discounting, not a price hike.
I built a recurring roster of ten webinars addressing real customer problems, owning topic research, offer design, funnel, nurture sequence, landing pages, and ad campaigns. Cold traffic came via Meta. Warm traffic, people who'd engaged before but hadn't converted, came via WhatsApp and IVR. It now outperforms standard paid channels on conversion.
Using Claude, Bolt, Zapier, Pabbly, Netlify, and Supabase, I automated lead uploads that used to be manual, and built an automatic WhatsApp welcome and feedback flow that fires the moment a product demo ends. I also built a real-time sales leaderboard so reps could see standings live instead of waiting for an end-of-day spreadsheet, which turned out to matter more for morale than I expected going in.
Three direct reports now own Social Media, ORM, and WhatsApp Marketing respectively. My job is less "approve everything" and more "make sure the three channels are telling the same story," since it's easy for a social post, a WhatsApp broadcast, and an ORM response to a review to quietly contradict each other if nobody's watching the seams.
Not vanity reach. Cost per qualified lead by channel, webinar attendee-to-sale rate, and AOV trend line. If a channel's CPQL creeps up two weeks running, it gets a creative refresh or a budget cut before it gets a shrug.
The same rough sequence shows up in both case studies above.
Audit the funnel and the spend first. Most "growth problems" are one specific leak wearing a disguise.
Fix the system, not the symptom. Landing pages, attribution, targeting, positioning.
Ship experiments fast, kill what doesn't work quickly, and say so honestly when something flops.
Double down on what's proven, automate the boring parts, and hand off a system, not a dependency on me.
Product teardowns and GTM strategy, done for competitions, or because a problem looked too interesting to leave alone.
Tide's loan application flow, rebuilt. Personas, competitor research and wireframes, up against 270+ senior PMs worldwide.
View deck →A 36-participant research study behind a full AOV improvement roadmap for Zepto.
View deck →Built on the idea that the real competitor for course-buyer attention isn't another bootcamp. It's ChatGPT.
View deck →A South India expansion proposal: market nuance, personas, funnel design and channel budgets.
View deck →I also occasionally write unsolicited growth theses for companies I find interesting. Here's one I put together for MyOperator, entirely unprompted.
Four working products, shipped end to end with AI-assisted development.
A full recreation of Classplus's marketing site, pricing tiers, feature grid and all. Built solo, front to back.
View site →The registration funnel behind the micro-webinar acquisition channel.
View site →A creator-content platform: sessions, masterclasses, ambassador storytelling.
View site →An internal attendance and activity-tracking dashboard, complete with real login. Built end to end.
View site →Four products, zero engineering tickets opened. That's not a boast, that's just what "AI-assisted development" is supposed to mean.
& many more.
Freelance projects, full time roles, or you just want to argue about whether attribution models are lying to you. All welcome.