# Sudeep Kumar, Product Design Lead I'm Sudeep, a design lead who builds. I make hard-to-understand products easy to use and trust, most recently at Carta. Now designing for humans and the AI agents working for them. ## Facts - Name: Sudeep Kumar - Role: Product Design Lead - Experience: 10+ years - Location: Canada - Open to: Full-time staff and senior IC product design roles. US and Canada first, open globally. Remote or hybrid - Most recent: Carta, 2019 to 2026 - Email: hello@sudeep.design - Website: https://sudeep.design - Linkedin: https://linkedin.com/in/sudeepkumarg - Twitter: https://twitter.com/heysudeep_ - Github: https://github.com/sudeepkumarg - Resume: https://sudeep.design/Sudeep-Kumar-Resume.pdf ## About I’m a Product Design Lead with 10+ years of experience across fintech and consumer products. I’m at my best with complex systems: the kind where the work is making something powerful feel simple, intuitive, and trustworthy. ### Background I went to school for advertising, marketing, and design, and started out in agencies. That’s where I learned to make an idea land. I still do that, just with products now. ### Getting into design Software design pulled me in because the hard part is rarely just the UI. I’m drawn to the model underneath: the rules, decisions, and interactions that make a new way of doing something feel obvious. ### Now I’m based in Canada. Most of my attention right now goes to AI and what it does to the shape of a product, mostly the unglamorous half: what to build, and how someone knows they can trust it. Voice most of all, where there’s no screen to fall back on. The rest is spread across travel, badminton, and F1 on Sunday mornings. ## How I work The way I design and build is changing as fast as the tools are. These are the principles that have held up. ### Think with AI, decide myself LLMs widen the option space early. The taste, framing, and final call stay mine. ### Prototype to decide A working prototype answers questions a mockup can't. I build early to make the right call, then ship with conviction. ### Hide the complexity The system absorbs operational complexity so the user never has to see it. ### Stay until the metric moves Launch only counts if behavior changes. I stay on it until the number moves. Current stack: Claude, Codex, Cursor, Figma, GitHub, Xcode. ## Experience ### Carta, Product Design Lead, 2019 — 2026 Led design for Carta's mobile suite and Global KYC: evolved Carta mobile to help employees manage their equity, launched Carta Carry (0→1) to help GPs manage funds on the go, and scaled KYC operations across 8 jurisdictions. ### Kik, Product Designer, 2018 — 2019 Shaped how millions of users discover and connect around shared interests on Kik messenger. Worked across safety, personalization, growth, and revenue teams. Acqui-hired by Carta in 2019. ### Stagwell Tech, Product Designer, 2017 — 2018 Designed consumer products spanning ride-sharing apps and video streaming platforms across web and mobile for the Canadian and Latin American market. Moved from Hubub acquisition. ## Selected work Each study in full, in the site's order. Locked studies show a public summary; email hello@sudeep.design for access. Every study also exists on its own at https://sudeep.design/project/.md. ### Carta Carry Carta · Design lead · 2026 · iOS and Android · Shipped #### Overview **Running a fund only worked from a desk.** Approving a capital call, releasing a wire, checking who has paid: none of it hard, all of it waited for a laptop. Carta Carry puts the fund in the GP's pocket. I led design from a deliberately rough first experiment with thirty GPs to a shipped app in four months. Month-over-month retention of GPs went from 35% at launch to 65% by early 2026, and it holds a 4.9★ rating on the App Store. ![The home screen. What’s active, what needs you. [raw]](https://sudeep.design/notion-images/dd3fee99f011bdf4aa8fff583b4ebdc4-raw-poster.jpg) Video: The home screen. What’s active, what needs you. [raw] (https://sudeep.design/notion-images/dd3fee99f011bdf4aa8fff583b4ebdc4-raw-1080.mp4) #### Where it started Carry didn't begin as a mobile project. The brief was speed: fund admin requests on the web took too long, and we were looking for ways to shorten them. The lead came from Carta Mobile. It was built for employees managing their equity and gave investors very little, fund information and performance, yet the data showed them coming back to it anyway, just to look. Nobody was running a fund from a phone. But they were already checking on one there, and that was a habit worth building on. So the question changed: not how to make the web faster, but what fund admin would look like if it started from the phone. Since nobody knew which parts belonged there, I kept the first version deliberately unstructured. Choose a request, and it opens a thread with the fund admin team. A polished guess is harder to walk back than a plain one, so the first release was a listening tool. We put it in front of about thirty beta GPs and let the requests tell us what the product was.   #### The design challenge A pattern showed up in the threads. Some requests came back with the same shape every time: capital calls, management fees, expenses. The routine machinery of a fund. The rest stayed one-off and high-stakes, the kind that needs a person. That split was the whole insight. The question was never how to automate fund admin. It was which parts not to automate. The routine wanted a fast, structured flow. The messy wanted a human, close by. Fund admin isn't complicated the way chess is complicated. It's complicated the way moving house is complicated: lots of pieces, lots of people, hard deadlines, real money. The job was to make the repeatable parts effortless on a six-inch screen without flattening the complexity or hiding what a GP needs in order to decide. The first instinct was a bot. A GP types a request, something answers. I explored it early and dropped it: a bot needs the full context of a fund to answer well, and in fund admin a confident wrong answer costs real money. So the request went to a person instead, and the routine got a structured flow around it. The name changed with the product. Concierge described the request line. Carry described the fund: carried interest, and the work of carrying it.   #### What it grew into Four taps, anywhere, and beside them a thread for anything that needs a person. The fast path and the human path, side by side. High-stakes decisions on a phone mean earning trust at every step. Nobody approves a $1M wire from a cab if the interface feels unsure of itself. The visual language, the copy and the confirmation flows all had to feel solid, and anything irreversible got the most time: on a wire approval, the confirmation screen and its copy were the last things I let go of. Over time Carry became where GPs handle the mechanics of the fund: who has contributed to a capital call, which wires to approve, where distributions stand, how the portfolio is doing. Each release added something. The through-line held: make the complex manageable, and keep the right action within a few taps.   #### Outcome - Month-over-month retention of active GPs went from 35% in the launch month to 65% by early 2026. Monthly active users grew 40% year over year. - The usage came from outside the office: airports, travel, hospitals. Approvals moved to mobile. - Carry became the companion to the web. Heavy workflows kicked off at a desk; status checks and approvals happened on the go. *Internal Carta metrics. Totals are shared credit with the team.* #### What I learned Carry got better when I stopped treating mobile as a constraint and started treating it as a different kind of focus. Desktop Carta answers "what is the state of my fund." Carry answers "what needs me right now." And the way to find that out wasn't to design it. It was to ship something bare enough that thirty GPs would show us. Canonical: https://sudeep.design/project/carta-carry ### Carta KYC — Analyst System Carta · Design lead · 2026 · Web · Shipped - Access: password protected. Email hello@sudeep.design to request it. - Outcome: An AI agent clears the clean cases so analysts get the hard ones Before an investor can put money into a fund, a compliance analyst has to confirm they are who they say they are and safe to let in. At Carta that was manual, one case at a time, with the analyst holding a dozen browser tabs together: vendor results in one, a search in another, an address checked by hand, screenshots and notes. It couldn't scale, and it didn't cover every region Carta operated in. I led design on the system analysts work in. First, bring the evidence onto one screen, deciding what a decision actually needs and leaving the rest out. Then, with Carta's ML team, a disposition agent that scores each case and clears the clean ones, so analysts spend their attention on the hard ones. The design problem was trust, not automation: every agent call shows its score, its reasoning and its evidence, and stays reversible. A fast first-pass colleague, not a black box. It shipped and is in daily use; cases that took hours now take minutes, and most never reach a person. The full study shows the analyst workspace, the evidence model and the agent's disposition UI. It is password protected because it depicts internal compliance tooling; email me for access. Canonical: https://sudeep.design/project/kyc-ai ### Carta KYC — GP Experience Carta · Design lead · 2026 · Web · Shipped - Access: password protected. Email hello@sudeep.design to request it. - Outcome: Investor verification GPs run themselves A GP raising a fund has to verify every investor before taking their money. At Carta that was a service, and not even one service: closings had a flow, onboarding another, fund admin its own, and the rest ran on email. GPs couldn't see where anyone stood, LPs got asked the same questions twice, and the one useful output, the risk score, never reached the customer. I led design on turning it into a product, in steps that transfer trust one at a time. First the process: a dashboard where a GP invites investors, watches checks move and exports the proof. Underneath it, one adaptive form across eight jurisdictions and seven entity types, so an LP is asked only what applies to them; twenty minutes became two or three. Then the judgment: the pipeline's internal states translated into five a GP can read, and the risk score put in front of the person who owns the fund. Next, designed and in progress, the decision itself, with an agent drafting the audit record for the GP to confirm. The full study shows the dashboard, the form system and the decision surface. It is password protected because it depicts customer-facing compliance product in detail; email me for access. Canonical: https://sudeep.design/project/kyc-gp ### Carta Mobile Carta · Design lead · 2023 · iOS and Android · Shipped #### Overview **Most people's first encounter with equity is confusing.** You join a startup, sign some paperwork, and somewhere in there you own a small, theoretical slice of the company. Carta Mobile is where employees at thousands of companies find out what that means and act on it: accept the grant, watch it vest, exercise when it's time. I supported the app for seven years, from a read-only window into Carta to a place people move real money. It holds a 4.8★ rating on the App Store.    #### Where it started I joined Carta through its acquisition of the Kik team, to build the mobile design practice. Carta's web product is built for admins: dense, powerful, oriented around the people running the cap table. The mobile app served the other side, and when I started it was a window you looked through. You could see your grants. You couldn't do much about them. The people on that side don't log in every day. The employee checking vesting on the train, the early hire wondering what the shares might be worth, the ex-employee deciding whether to exercise before a deadline. They log in when something is happening, and every session carries some weight.  #### The design challenge Equity is one of those topics where the complexity *is* the product. Vesting schedules, exercise windows, strike prices, fair market value: you can't flatten it into a single number without losing something important, and you can't dump a spreadsheet onto a phone and call it done. The job was finding the line between clarity and completeness for people checking between meetings, not at a desk with a calculator. Exercising was where that line mattered most. Employees who could exercise often didn't. Plenty didn't realise that an option is only the right to buy shares, and that exercising is the purchase itself: real money out, and a tax bill that arrived as a surprise. The fear wasn't of the money. It was of not understanding the money. So instead of a shorter form, we built the understanding into the flow: plain explanations at the point the question comes up, and a simulator that models the outcome and the taxes before anyone commits a dollar.  #### What it grew into Over time the app went from a read-only window into Carta to a place where people do things: accept grants, simulate an exercise, exercise, track holdings across currencies. Each release closed a gap between what the phone showed and the decision someone needed to make. The through-line stayed the same: meet people where they are, not where a finance team assumes they are.   #### Outcome - Monthly exercise volume went from roughly $5M to between $10M and $20M within three to four months of the new flow shipping. Completion and drop-off on the flows I owned improved directly. - Tens of millions of dollars in options now get exercised from a phone, across thousands of companies. - 4.8★ on the App Store. *Internal Carta metrics. Volume totals are shared credit with the team.* #### What I learned Trust is load-bearing in a financial product. When someone is looking at a number that is a real chunk of their net worth, every detail matters: the alignment of decimal points, the copy on the confirmation button, the freshness of the data. People feel uncertainty in an interface before they can name it, and here uncertainty is what makes someone close the app and open a laptop. The simulator taught me the cheapest way to earn that trust: let people rehearse the decision before they make it. Canonical: https://sudeep.design/project/carta-mobile ### Even Independent · Designer and developer · 2026 · iOS · Built, not shipped #### Overview **Expense apps bury the one thing you came for.** Even is a bill splitter: snap a receipt, see the split, settle up. It's also an experiment: how far a designer can take a real product alone, with AI writing the code I can't. Not a prototype. The whole thing: receipt parsing, haptics, accessibility, the App Store bar. The app I wanted to use and the question I wanted to answer, wearing the same coat.   #### **Where it started** A group trip. Six people, a stack of receipts, and someone with a calculator app doing math nobody wanted to double-check. By the end of the week the group chat was half jokes, half spreadsheets. I wanted the opposite. Point your camera at the receipt, tell Even who was there, done. And I had a second itch. Splitting a bill is a small, well-shaped problem. Small enough to actually finish, which made it the right place to ask a bigger one: with AI, how much of a real product can one designer carry?  #### The design challenge Two problems, stacked. The product one: expense apps bury the thing you came for, under groups, friends, feeds, and nudges to upgrade. I wanted the hero flow to be one confident line, scan then assign then settle, and everything else to be scaffolding around it. The scan had to feel like magic without pretending it's infallible, because receipts are messy. So the results screen shows what it found, lets you fix anything, and never makes you start over. The build one: I'm a designer, and AI would write the Swift. That sounds like a shortcut, and in places it is. But it moves the hard part rather than removing it. AI will write you a settlement animation in thirty seconds. It will not tell you the animation is worth two more days, or notice that a green amount and a red amount look identical to someone who's colour blind. It builds the easy version of every screen and never mentions there was a better one. That gap turned out to be the whole job. The slow part was being the one who says no, do it again, this isn't it yet. Every amount ended up carrying a symbol and an arrow, not just a colour. Every animation got a Reduce Motion fallback. Every tap, toggle, and settlement got its own haptic. None of it something AI volunteered. All of it design.  #### What it grew into Even became a small, opinionated thing. Monochromatic green. SF Pro Rounded. No ads, no limits, no paywall waiting to spring. The opposite of the app it was a reaction to. The part I'm proudest of is the smallest. When a group finally squares up, the equals sign pulses, a line fades in that reads *you're even, as it should be*, a bell plays, and a haptic crescendo runs under your thumb. It lasts about a second and is wildly disproportionate to its function. That is the point. The whole reason to split a bill is to get back to being even with your friends, and that moment deserved more than a toast. AI built every frame of it. It would never have argued for it.   #### Outcome - A complete iOS app: on-device receipt scanning, splits by item, percentage or exact amount, settlement, a haptic on every action, Reduce Motion fallbacks, amounts that read without colour. Built to the App Store bar. - Not shipped, by choice. Splitting a bill belongs next to the money, in the wallet you already carry, not in a standalone app you have to talk your friends into downloading. The platforms will get there; they're closer than they look. I'd rather be early and right about where this should live than ship something the operating system is about to make redundant. #### What I learned AI closes the gap between designing a thing and building it. For a designer working alone, that is new, and freeing. I built things I could never have built by hand. What it doesn't close is the gap between building it and building it well. That gap is taste, standards, and the willingness to redo work nobody asked you to redo. It is still, exactly, the design job. The tool got faster. The judgment didn't move. I didn't build Even to win a market. I built it to find out what's left of the design job when the building gets easy. Turns out it's most of it. Canonical: https://sudeep.design/project/even-app ### Kik — Privacy Controls Kik · Product designer · 2019 · iOS and Android · Shipped #### Overview **Joining a public group on Kik handed every member a direct line to your DMs.** Public groups are rooms anyone can join, and one of the main ways people find each other on the messenger. For some, an open inbox is the point. For others it is the creep at the party following you to the kitchen. My job was to let people shut that door without leaving the room. We shipped it, then reshaped it when the data said we had built it the wrong way round.  #### Where it started Users loved the groups. They did not like the side effect: a stranger now had a direct line to them. The reviews said it plainly. This was not an annoyance. It was a reason people felt unsafe, and unsafe people leave. #### The design challenge Safety settings have a trap: the people who need them most never go hunting for them. So two questions. Turn DMs off everywhere at once, or group by group? And does the control live in global Privacy, or in the group, where the discomfort actually happens? The global switch looked simpler, so we shipped that first.  Then we asked. The feature was wanted: 83% liked having the control. But the all-or-nothing switch was the wrong shape. 89% wanted to turn DMs on or off per group, the highest number in the study. It makes sense. You might chat openly in your hiking group and stay sealed off in a 300-person channel. V1 gave people a light switch for the whole house when they wanted one for each room.  #### What it grew into V2 moved the control into the group. A "Receive Direct Messages" toggle in the group info screen, next to notifications, no menu-diving. Open where you want to be open. Closed where you do not. And it shows on your profile: with messages off, the door reads as shut.   #### Outcome - 83% of people surveyed liked having the control. 89% wanted it per group. V2 shipped per group. - "It keeps the creeps away." A user, 18, in the reviews. #### What I learned Safety is not one wall you build once. It is a set of small, situational choices, and the best thing you can do is hand people the controls and trust them to know their own rooms. It made Kik feel better to be on. Which was the whole job. Canonical: https://sudeep.design/project/kik-mobile ## Experiments Interactive studies at https://sudeep.design/experiments. They are things you use rather than read, so this file describes them. ### Voice Order You already know what you want. The app makes you browse for it anyway. Try it: Tap the mic, then place the order. https://sudeep.design/experiments/voice-order ### Slash Commands Act on the text above the cursor without selecting it, opening a toolbar, or leaving the keyboard. Try it: Type / and pick a command. https://sudeep.design/experiments/slash-commands ### Overnight An agent acted on your behalf all night. Now it has to account for all five, including the one it isn’t sure about. Try it: Open the one that needs a decision. https://sudeep.design/experiments/overnight ### SwiftUI Fluid Motion A skill that makes Claude and GPT write better iOS animations. Try it: Download the skill, or copy the install line. https://sudeep.design/experiments/swiftui-fluid-motion ## Contact Email hello@sudeep.design. I reply to real messages. Locked case studies are available on request. This text may be summarized and quoted with attribution to sudeep.design. Link to the canonical page for the current version.