I'm Roshni, a product leader working across product-led growth, risk & compliance, and the platforms that connect them.
How I work
I work across the whole system: product-led growth, risk & compliance, and the platforms that connect them. Each part reinforces the others instead of fighting them.
Growth
Product-led growth, onboarding, activation, PLG motion design
Risk & Compliance
KYB, KYC, AML, fraud prevention, decisioning platforms
Platforms
Internal tooling, operational platforms, case management
Leadership
Cross-functional teams across Product, Eng, Data, Ops, Compliance
The product challenge
Most companies treat compliance as a gate. Growth teams optimise for conversion. Risk teams optimise for control. Operations absorbs the inefficiencies in between.
The question I ask is not how to choose between growth and compliance. It is how to design systems that make both possible at the same time.
"Compliance should be an asset, not a bottleneck."
Selected work
Six initiatives I owned end to end, plus two on how I think when the answer isn't obvious.
Phishing-led account takeovers were 73% of our fraud losses, and the defence was scattered across squads. I priced every gap in euros, built the three-lever strategy, and won a five-month tiger-team mandate.
View project →Top-of-funnel was improving but activation wasn't following. I pushed past aggregate conversion into weekly cohorts, caught the KYC completion slide hiding in the averages, and refocused six teams on one constraint.
View project →Sales couldn't show the real product without bespoke setup, and the guided tour showed a product that no longer existed. I drove a guard-based, self-serve demo of the real app, built ground-up with AI in two months.
View project →Our KYB tooling ran on a backoffice that couldn't scale. I led the buy-vs-build analysis to a sharper answer than either extreme: buy the commoditised verification automations, build the case management that differentiates.
View project →KYC feedback loops looked like an ops workflow. In reality they were a silent conversion killer. I reframed loops as a product metric, instrumented the causes, and cut avoidable loops without touching a single control.
View project →Signup is the cheapest place to win: every point gained multiplies downstream. Disciplined experiments on the first two steps of a regulated funnel moved both metrics, hard, without poisoning cohort quality.
View project →Four questions, every time: differentiation, true cost of ownership, regulatory surface, and speed to be right. Verification was a buy. The PLG demo was a build. Same rubric, opposite answers.
Read →I don't read about AI, I build with it. This portfolio is my reflection journal; Morsel is a food-tracking app where the model never invents a fact. Both feed straight back into the work.
Read →How I think
These are not process rules. They are design decisions I make repeatedly, in different products, across different teams.
01
Trust is earned, not assumed
Not every user needs the same level of scrutiny. The cost of treating everyone as high-risk is paid in conversion and customer experience. Apply verification proportionally to the actual risk in front of you.
02
Friction is a design decision, not a compliance requirement
Every step you add to an onboarding or verification flow has a cost. The question is not whether friction exists but whether it is placed at the right moment for the right reason.
03
The operating model is part of the product
Compliance workflows that live in spreadsheets and email threads are not compliant. Policies need to be configurable. Decisions need to be auditable. Controls need to be measurable.
04
Automate the volume, protect the judgement
Analysts should be investigating exceptions, not processing queues. Automation is not about replacing humans. It is about reserving human expertise for the cases where it actually creates value.
05
Auditability is a product feature
Every decision should be explainable. Every control should be traceable. The audit trail is not added at the end because a regulator asked for it. It is designed in from the start.
06
Good metrics span both sides of the trade-off
Optimising for conversion at the expense of fraud losses is not success. Optimising for risk controls at the expense of activation is not either. The dashboard needs to show both at once.
What good looks like
Optimising one dimension at the expense of the others is not a win. These four groups define the full picture of a healthy platform.
Growth
Activation rate
Conversion rate
Time to first value
Risk
Fraud losses
Regulatory incidents
False positives
Operations
Straight-through rate
Analyst productivity
Review queue size
Experience
Verification completion
Customer effort score
Time to approval
About
I'm a Product Lead based in Stockholm, specialising in the intersection of growth and trust. I've spent years building and scaling fraud, AML, KYB/KYC, and onboarding platforms used by millions of customers across fintech. My edge is making compliance a competitive advantage: friction becomes conversion, and risk becomes a product feature.
I grew up in Chennai, and somewhere between India and Sweden I developed strong opinions about product, cold-weather hiking, and the right ratio of oat milk in a flat white. I have done Spartan races, chased the Northern Lights in Abisko, and I am convinced that fika is one of the better cultural inventions of the 20th century. I share my apartment with a dog and a cat who have very different opinions about that.
I am curious by default, collaborative by nature, and outcome-driven above everything else.
"Compliance should be an asset, not a bottleneck."
Connect on LinkedIn4+
Teams led simultaneously
19+
Specialists across disciplines
€625K
quantified
Preventable fraud losses priced & funded
+59%
Conversion lift from growth experiments
Based in Stockholm. Originally from Chennai.
Hiking trails, Spartan races, Northern Lights in Abisko, fika, dog and cat parent.