Freelance
Give me the job.
For smaller, defined jobs. You tell me the outcome, I scope it, do it and hand it over.
Priced per job. Three client slots open.
Scope a job, see the cost →I find that work, build the AI systems that take it over, and keep a person on every call. Your people get their week back for the work that grows revenue.
Taking 3 freelance clients · full-time from May 2027Companies I have worked with since 2015

I started in mechanical engineering and moved to marketing because I found the translation problem more interesting: why good products fail to connect with the people who need them.
Since then I have built marketing functions from scratch, worked solo, built teams and co-managed a budget that went from ₹50 lakh to ₹5–6 Cr a year, in hard sectors like BFSI, telecom and government. Today I replace redundant marketing labour with AI-augmented systems, so production scales without scaling headcount.
I am an AI-first marketer. I accelerate business outcomes with AI-assisted marketing systems. It comes down to three things.
I map the work your team repeats every week, then build the AI pipeline that does it: scripts, edits, ad variants, translations, reports. A person still sets the brief and signs off every asset.
Positioning, story and creative direction that make buyers remember you before they need you. I was the brand owner at Gnani.ai and took EdNxt.ai from zero to a launched brand.
Paid leads stop the day the money stops. I build channels you own on YouTube, LinkedIn, Instagram and search that keep bringing leads in long after they are built.
An engineer who moved into marketing because the interesting problem was never building the product. It was why good products fail to reach the people who need them.
A budget is not a result. What matters is what each rupee bought at every stage of the funnel.
a year for marketing and sales, co-handled with the VP of Marketing, across JAIN Online, EdNxt.ai and about 20 colleges of the JAIN Trust, formerly USDC Global. Before that, ₹50L grown to ₹5–6 Cr a year at Gnani.ai.
Reach · cost per view~₹0.05a view. About 20 million views on about ₹10 lakh for a JAIN Online awareness campaign at JAIN Trust, formerly USDC Global.
Leads · volume against paid2.2×the usual Meta and Google lead volume from one organic push at JAIN Online.
Pipeline · influenced₹40 Cr+enterprise pipeline influenced by 200+ sales and pitch decks across government, enterprise and event partnerships.
Efficiency · production cycle−60%shorter with AI workflows, so the same budget buys more output.
Capital · investment round$4Mfrom Info Edge. I built the collateral and ran the communications.
I report on CAC, cost per lead, MQL-to-SQL conversion and payback period. Those depend on your category, so ask me for the breakdown that matches yours.
Building got cheap. Getting noticed did not. Every competitor can now ship more posts, more ads and more pages than they could last year.
The same two or three ads carry the account until they burn out. The fix is a steady supply of new creative, not new audiences.
What I do about it ↻What I do about itThe AI creative engine: 12 to 20 new ad concepts a month, before fatigue sets in.See how → Generic AI work gets ignored.Buyers can tell when nobody directed the work, and they scroll on.
What I do about it ↻What I do about itA person directs and approves every asset. If it looks generic, it does not ship.See how → Rented leads vanish.Performance marketing is a tap. Close it and the leads stop the same week.
What I do about it ↻What I do about itOrganic lead engines on Instagram and YouTube. One push at JAIN Online beat paid by 120%.See how → Small teams are out-shipping big ones.One owner with an AI pipeline now does what used to need a department.
What I do about it ↻What I do about itI build that pipeline in your accounts, so your team is the one out-shipping.See how →
Everyone tried AI for content. Most of it landed in the same place: more volume, less attention. The teams that got through are the ones that put a person in charge of the machine.
My reading of the market, drawn as a maturity curve. Not a published chart.
This is the difference between renting attention and owning it. Paid gives you leads this week. A content engine takes longer to build, then keeps paying.
An illustration of the pattern, not data from a campaign.
Without AI, cost rises with every asset you add, because every asset needs hands. With a pipeline, the cost flattens after it is built.
An illustration of the shape, not measured data.
Freelance
For smaller, defined jobs. You tell me the outcome, I scope it, do it and hand it over.
Priced per job. Three client slots open.
Scope a job, see the cost →
Hire me
One owner for marketing, with an AI pipeline doing the volume. For founders who want one person answerable for the outcome, not five vendors to manage.
Full-time from May 2027.
See what you would be hiring →
Mentorship
One-to-one sessions for marketers and founders who want to build these systems themselves.
Charged per hour. The rate depends on the effort the problem needs.
Ask about mentorship →Eight or more Indian and international events in about 18 months at Gnani.ai. Logistics, vendors, collateral and on-ground execution, then the follow-up that turns a badge scan into a meeting.

A 221 m² showcase for Gnani.ai: the stand, the collateral and the team on the floor. Other faces are blurred.
“What a show you guys pulled.. kudos”
Pushpa Sahay, Director, Strategic Accounts, on LinkedIn
“Spectacular work Saptarshi Sarkar … Your perseverance is commendable”
Jayakrishnan P V, Enterprise AI Transformation Leader, on LinkedIn

Voxi, the mascot I designed, at the India AI Impact Summit 2026. Other faces are blurred.
“Voxi was the highlight of our booth Saptarshi Sarkar kudos”
Pushpa Sahay, Director – Strategic Accounts, on LinkedIn


The Gnani.ai stand mid-build at Bharat Mandapam, before the summit began. Then the same stand, finished. Built in about 72 hours before the summit. Other faces are blurred.

International trade show.

Government and defence buyers.

Customer-contact industry show.

Global capability centre leaders.

Interview sets for JAIN Online: lights, sofas, two cameras and a host.
An AI marketer is someone who uses AI to reach the same marketing goals. The goal does not change. The route does. Pick a job and compare the two routes.
The job: A month of ads that do not burn out
Days per ad. One or two ads carry the account until they tire.
12 to 20 new concepts a month, before fatigue sets in.
Sample layouts to show the difference in volume, not real campaigns.
The job: Outreach that lands in the buyer’s morning
Tied to one person’s working hours and memory.
Runs around the clock. Nobody waits up for a time zone.
The job: A deck for tomorrow’s enterprise meeting
Days. The request often outlives the opportunity.
Hours, not days. This way of working grew out of building 200+ decks at Gnani.ai.
The job: A product film without an agency
Weeks, an agency invoice and a single version.
Script to finished asset without editors or agencies: 60% less time.
The job: Numbers the founder can act on
Monthly, manual and already stale when it lands.
Weekly, and it feeds the next decision.
Production that took days takes hours. At Gnani.ai the production cycle fell by 60%.
MoneyOutput grows without adding headcount, editors or agency retainers.
ConvenienceThe pipeline runs day and night. Nobody waits for a shift, a slot or a time zone.
An agent is an AI worker with one job, the tools for it and rules it cannot break. I build them as a team, so the work passes from one specialist to the next and comes back checked.
Brand owner for a conversational-AI company selling to BFSI, telecom and government. 200+ decks, global events and the $4M Info Edge round.
See the numbers → 2026Instagram and YouTube turned into a lead source. One push beat the usual Meta and Google volume by 120%.
See the curve → ednxt0 to 1Positioning, product story, creative direction and go-to-market for IIT, IIM and NIT programmes.
See brand lift → YouTube · RoadHook TouringAI filmmakingThe Sikkim Diaries: solo on a 200cc, cut by directing Claude Code and DaVinci Resolve. The proof that the method works.
Watch on YouTube →Bazaar is my marketing blog: how marketing and sales work after AI, with sources. Start with these.
No. I take over the repetitive production your team should not be doing by hand: drafts, variants, edits, reports. Your people keep the judgement, the relationships and the decisions. Where a team is small, the pipeline lets it do the work of a much larger one.
It should not. A person sets the angle and approves every asset, and anything that reads as generic goes back. That check is the reason the output holds up. AI is disclosed wherever platforms or ASCI rules call for it.
Access to the accounts the work runs in, whatever brand and product material exists, and one person who can say yes or no. I begin by mapping the work your team repeats every week, because that is where the first system goes.
You do. Everything is built in your accounts and documented, so it keeps running after I leave and anyone on your team can operate it.
It depends on the system. Production systems, such as decks, ads and video, change the pace in the first weeks. Content engines take longer, because channels you own need time to build, and that is also why they keep paying after the budget stops. I will tell you which kind you are buying before we start.
Freelance work is priced per job, and the brief page gives you a figure for your scope. Mentorship is charged per hour, and the rate depends on the effort the problem needs. Full-time, I am available from May 2027.
Enterprise and conversational AI, higher education and ed-tech, and mobility and manufacturing, with buyers in BFSI, telecom and government. The systems carry across industries. The buyer research is what changes.
Yes. I have run marketing solo, led a team of four and managed vendors. The pipeline can sit beside an agency and take the volume work, or it can replace the parts you no longer want to pay for.
Yes. The same approach works for sales: agents for account research, first-draft outreach, follow-ups and battle cards, with a reviewer in the loop. I have spent years building sales enablement by hand, so I know which parts are safe to hand to an agent.
Usually not. Most of it runs on a small set of tools you may already have, and I will tell you what each one is for before you pay for anything.
Yes. That is what the mentorship is for, and every system I build comes with documentation, so your team is not dependent on me.
Ask. If it sits next to something here, such as an internal tool, a new channel or a launch, I can probably do it, and I will say plainly if I cannot.
What is eating your team’s week?
One brief a week becomes posts, reels and articles, with a person approving each one before it ships.