Anisha Kumar
"Builds 0→1 AI platforms where the hard part isn't shipping the model — it's making the output trustworthy at enterprise scale."
Good products start from friction, not inspiration — a real pain point, an inefficiency someone's already working around by hand. What I'm actually optimizing for is whether that friction disappears: does the metric move, does someone's day get measurably easier. That's the only definition of "good" I trust.
I like living in the details of how something actually works — the schema, the edge case, the architecture — but I check that against the bigger picture just as often, so the two stay honest together. I default to explaining the "why" behind every decision, not just the "what": years of TA'ing taught me that if I can't explain something at three different altitudes to three different people, I don't understand it well enough myself. And I'd rather build 0→1 with a small team wearing five hats than hand off a spec and wait — there's nothing like watching the thing you built actually stop the problem you set out to solve.
Studio
In productionWalmart's internal AI analytics platform — turns natural-language prompts into governed business intelligence for business teams, engineers, and product managers across asset protection, finance, people, merchandising, sustainability, and governance.
Turned a tool that guessed at data sources into one that answers a single sentence with a governed report — adoption grew 4.5× in 3 months, with 82% of users coming back.
When I took over Studio, it was a proof of concept, not a product — something that worked in a demo but wasn't ready to hand to a business team, let alone get adoption. Underneath, it was guessing: which data source you meant, which model to use, based on whatever you happened to have access to. Guess wrong and you got the wrong table, ungoverned numbers, or a chart that didn't match the question. Getting to that first guess took six setup steps and a separate login for every data source, and once you had something, there was no way to hand it to anyone else — sharing a dashboard meant copying the raw JSON behind it into a message.
I didn't start by adding features — I started by deleting the six steps between opening Studio and a first prompt, since they weren't protecting anything, just leftover scaffolding from an earlier version. Then I went after the guessing itself. The fix wasn't a better guess; it was giving the system something real to reason with — metadata about the data: what's certified, what domain it belongs to, who uses it, what access group you'd need. I designed an orchestration layer that reads all of that against your own session history to figure out what you're actually asking for, instead of what you happen to have a login for.
None of this matters if people can't trust what comes out the other end. Alongside the generation work, I built what makes Studio's output something you can act on without checking it by hand: a skill that validates a generated report against its real source of truth, a skill that enforces Walmart's design system so nothing generated looks out of place, and a skill that understands Walmart's retail calendar so a "this quarter" question doesn't quietly answer the wrong quarter. None of it is flashy. All of it is the difference between a dashboard someone glances at and one they build a decision on.
The single most requested feature, from analysts up through VPs, was real-time collaboration — and I didn't build it. I bet that a fast, trustworthy single-player experience mattered more than a slower multiplayer one, and held the people asking for it with publishing, version history, and a chat that already understands every number on a dashboard instead. I pressure-tested that bet with competitive analysis against Power BI and Walmart's other internal tools, then used it to shape how Studio got pitched to leadership. And rather than build every capability alone, I mapped where Studio needed to plug into tools people already used — building the data-catalog and AI-platform integrations myself, and scoping the coding-agent and CLI work for later.
Getting to production wasn't a feature-complete moment, it was a compliance one. Security certification had stalled for reasons that had nothing to do with the product being ready. I pushed it forward, made sure production infrastructure was built and waiting before the approval landed instead of after, and personally connected the people who needed to sign off on each piece. Then I found the first users myself — cold-calling people stuck trying to build dashboards in other internal AI tools. Every one of them came back, and within three months ten more business teams had migrated onto Studio.
Next time, the administrative steps — the security certification, the approvals — start on day one, not whenever the build has room for them; they gated the timeline more than any of the technical work did. The bigger lesson was about who Studio is for. Most tools like this only solve for the person building the dashboard. I built for the person just trying to read one, too — and that view-only mode ended up being one of the most-used parts of Studio in trials, proof that the niche the existing tools left open mattered as much as the orchestration underneath it.
Risk Detection
In productionAn investigation platform covering every Walmart U.S. and Canada store, processing 600K high-risk line items weekly for 30 researchers across 7–8 teams — built with a 6-person engineering team spanning the US and India.
Built the investigation platform every Walmart store in the US and Canada now runs on, replacing a fully manual process — cutting investigation time up to 70% and turning 600K flagged items a week into decisions instead of backlog.
Before this platform existed, investigating a flagged item meant stitching together data from sources that were never designed to talk to each other. Investigators worked across email threads and spreadsheets, with no shared record of what had already been looked at, what had been actioned, or why. Nothing scaled. Every store, every week, the backlog grew.
I didn't start with the technology — I started by sitting with the researchers who'd spent years learning to spot patterns nobody had ever codified, and worked with them to turn that judgment into detection logic a system could act on. Before engineering committed to anything, I prototyped and validated the review flow myself in Figma and v0, so stakeholders could react to something concrete instead of a spec. What shipped pairs a model-generated signal with an investigator's review, so every decision they make feeds back in and sharpens the next one.
The bigger call was architectural: consolidate every fragmented source into one pipeline, instead of patching the spreadsheets people already had. That was a heavier lift up front, but patching would have left the real problem — the manual wrangling underneath the backlog — untouched.
The biggest shift for me was realizing how much more a clickable prototype does than a requirements doc. I could put something in front of stakeholders and watch them react to it in real time, iterate on the spot, and hand engineers something unambiguous instead of a spec they'd have to interpret. And there was something particularly satisfying about this one specifically — the research team had been doing every bit of it manually, store by store, with no real technical support before I joined. Watching that actually change was the whole point.
Intelligent Business Growth
Finance teams were planning off static assumptions that didn't hold. I replaced them with ML-driven forecasts across the planning workflow.
Connectivity
Four finance systems were duplicating the same work in isolation. I founded the initiative that unified them under one shared architecture.
Enterprise Reporting
Finance leadership's numbers lived in static monthly packets. I turned it into a live hub with KPIs, market context, and AI-generated analysis — then handed it off intact to move onto Studio.