Who's reading?

No password. Your answer changes what opens first and never hides anything, and it's forgotten when you close this tab.

No password. Forgotten when you close this tab.

Prefer a regular website? Open it hereRegular website

Abhishek Arun

Associate Product Manager at Internshala, based in Gurugram, India, originally from Kerala. Open to opportunities.

This portfolio is built as a desktop you log in to. A regular version of this site is also available.

Selected work

SpendFlow: Privacy-first personal finance PWA

Status: Live.

Problem: Personal finance apps want your bank login. The people who need the help most are exactly the ones unwilling to hand that over.

Insight: The friction was never the maths — it's the logging. And a finance app that never phones home removes the reason to hesitate.

Decision: Natural-language entry ("400 groceries"), with answers computed on-device. No account linking, no server-side ledger.

Kept: Spending data that never leaves the phone. Gave up: Automatic bank sync — every rival's lead feature.

Built on: Installable PWA with local-first storage; on-device parsing and querying, so spending data never leaves the device.

Afterward: Digital legacy planning

Status: Live.

Problem: When someone dies, their digital estate is a locked box nobody has a map to — accounts, subscriptions and assets scattered across services.

Insight: In a trust-first category, the thing you refuse to build is the product. A credential vault would have made security the whole conversation.

Decision: Index only. Afterward documents what exists and who to contact, and never stores what unlocks it.

Kept: Zero credentials stored, by design. Gave up: The password vault — the most requested feature.

Built on: Encrypted metadata index with no secret storage by design, plus a nominated-contact release flow.

Focus Docs: Minimal doc collaboration

Status: Live.

Problem: Doc tools bloat into everything-suites. Teams just want to write together without the distraction.

Insight: You beat bloated tools by subtraction, not by catching up. The differentiator is granular control with zero friction.

Decision: Cut comments, templates and integrations from the MVP. Shipped per-section access control as the only differentiator.

Kept: Per-section access control, done properly. Gave up: Comments, templates, integrations — and every feature table.

Built on: Section-level access-control model over a minimal real-time collaboration layer.

Tesseract: School timetable & substitution management

Status: In pilot this academic year with a school in Kerala.

Problem: School timetabling and daily substitutions break under scale — done by hand they eat hours and still conflict.

Insight: The hard part isn't the UI, it's the constraint solver underneath. Get that right and the rest follows.

Decision: Built the solver first, with almost no interface. Sequence the risk, not the visibility.

Kept: A solver that resolves real conflicts. Gave up: Anything demo-able, for months.

Built on: Constraint-satisfaction engine resolving timetable and substitution conflicts in real time, on a Firebase backend.

Experience

Associate Product Manager, Internshala, Jul 2026 to present. It's a recent move, so there's no shipped work to point to there yet.

Product Manager Intern, TruckHai, Feb 2026 to Apr 2026.

Product Management Intern, WebMobi, Nov 2025 to Jan 2026.

B.Tech, Computer Science, SRM IST, 2022 to 2026.

Teardowns

  • UPI for the next billion: Designing payments for users with low literacy, patchy networks and shared phones.
  • Redesigning LinkedIn for Gen Z: A generation that proves competence by showing work, not listing titles.
  • Food delivery economics in India: The per-order model can't win, so change the unit.
  • ChatGPT for students: The line between a learning tool and a cheating machine is a product decision.
  • School administration software: The Tesseract thesis — why I built it, and where it goes.
  • Indian Railways, digitized: At massive scale the deepest pain isn't the UI — it's the uncertainty.
  • AI-native SaaS: When agents do the work, the human shifts from operator to supervisor.
  • Spotify discovery: Recommendation solves "play me something." It doesn't solve "find me this."
  • What PMs must learn for 2030: The measurable skills are automating first. Bet on judgment.

Capabilities

Product: User research, Jobs to be done, Product strategy, Prioritization, Competitor analysis, MVP planning, Wireframing, User flows.

AI: LLM use cases, Prompt engineering, AI safety UX, Human in the loop, AI evaluation, Personalization, AI ideation, AI workflows, RAG basics.

Build: React, Firebase, HTML / CSS, Supabase, REST APIs, No-code tools.

Systems: System design, SQL, GitHub, Analytics.

Execution: PRD writing, Roadmapping, Agile / Scrum, Sprint planning, A/B testing, KPI definition, GTM strategy, Stakeholder comms.

Certificates

  • AI for Product Management, Pendo.io
  • Product Management Basics, Simplilearn
  • Introduction to JIRA, Simplilearn SkillUp
  • Networking Basics, Cisco
  • Rajya Puraskar, Scouts & Guides, Kerala

Contact

LinkedIn · Résumé