Sumeet Kumar · Senior Product Manager · Bengaluru
Seven years of product management across SaaS, CRM, marketplaces, and EdTech — five 0→1 launches, a founder chapter, and two production AI systems now serving twenty thousand learners. IIM Visakhapatnam MBA. Currently Senior PM at Simplilearn.
Highlights from every chapter — production AI systems at Simplilearn, AI learning tools at Edfora, revenue-critical CRM work at LeadSquared, two 0→1 SaaS products at Click Labs, and a founder run at Nazdik.
A Supervisor Router orchestrating four specialised agents — Learning Buddy, Project Buddy (B2B), Help & Support, and Context Switch — that classifies every learner query and holds sticky, multi-turn context across topic changes.
An LLM pipeline fused with a deterministic rule engine that watches every concurrent live class through a six-state session-health model — surfacing issues to operations teams in real time, before learners have to complain.
Two years building AI-assisted learning before it was fashionable — summarisation, adaptive models, and the analytics that tell instructors which learner needs them next.
Owned Forms & Processes in the Sales Workflow division — the revenue-critical path where every basis point of failure is money.
Designed, built, and launched Husky (freelancer marketplace) and Juggernaut ERP (HRMS, POS, WMS) from zero — then sold them to the world.
Built and launched a hyperlocal commerce marketplace (mobile + web) connecting local retailers and consumers — then wound it down against pre-set unit-economics kill criteria.
The unglamorous half of the current role at Simplilearn: making seven systems — LMS, cohort operations, web CMS, revenue, mobile, third-party LMS, and a free-learning funnel — agree about what a learner owns, sees, and pays.
One source of truth for learner access across four platforms, with conflict-resolution logic for shared entitlements and B2B extension rules.
End-to-end swap workflow across revenue, cohort ops, and CRM — three swap modes, dual-invoice accounting with GST treatment, automated entitlement actions.
Global-default + cluster-override pricing with geo rules, audit logging, and a backward-compatible migration across a live catalog.
Timezone/DST-aware scheduling, registration flows, Zoom webhook & attendance-sync migration, and a self-hosted tokenized calendar feed that retired a paid vendor.
Shipped the blended-learning experience on Android (iOS in development) with full lifecycle-event instrumentation from day one.
Integration requirements for IBM SkillsNetwork (SSO/LTI, score sync); university-partner delivery on Canvas; a certificate-platform build-vs-buy decision with a Month-9 readiness gate.
Breadth first — SaaS, CRM, a founder chapter — then a deliberate commitment to one thesis: AI-native learning products.
Two production AI systems (Arya, Control Centre) plus platform ownership across validity, pricing, revenue workflows, and live-class infrastructure.
AI summarisation and adaptive learning models; app redesign that lifted time-on-platform 90%; gamification driving +27% adoption, +20% retention; a cohort-analytics score surfacing at-risk learners for instructors.
Hyperlocal commerce marketplace — 500+ merchants, 10,000+ SKUs. Wound down on pre-set unit-economics criteria; the discipline stuck.
Forms & Processes on the CRM platform: payment failures 5% → <1%, pricing restructure +25% revenue, auth flows +12% registration.
Two 0→1 SaaS products (freelancer marketplace + ERP suite) with a matrixed team of 12 — USD 1M+ from 150+ global customers.
I'm a product manager who likes the whole stack of a problem — the agent prompt and the entitlement table it depends on. Most of my last four years have gone into one question: what does great AI-assisted learning actually look like in production, once you're past the demo and into vector stores, routing failures, cost curves, and ops teams at 2 a.m.?
Before that: an MBA from IIM Visakhapatnam (Finance & Analytics), an engineering degree, a CRM platform, an ERP suite, and one startup that taught me more per month than anything since.
I write occasionally about AI product management — routing design, agent metrics, and the honest versions of metric stories. Say hello if you're building in this space.