Jetson | Portfolio
Jetson SamuelGet in touch
Open to PM & APM roles

Jetson Samuel

Associate Product Manager at Promanage

Connecting the dots between users, business and tech.

See my work ↓ResumeLinkedIn ↗
Jetson Samuel, arms crossed, in a black blazer
500AI answers audited, 67 hallucinations caught
25PMs interviewed before I built Knit PM
33KPeople follow my Tamil and English channel
3 moResearch to public launch, as the only PM at an AI security startup
Experience

Where I've done the job

  1. ProManage, a Sulekha business · Chennai
    Current
    Associate Product Manager Intern Oct 2026 – now
    • Evaluated a production AI agent with about 500 real and edge-case questions, tracing every tool call to find where answers went wrong.
    • Proposed moving from black-box to grey-box testing, so each failure can be traced to retrieval, ranking or generation.
    • Designed a per-seat pricing model for a new support product.
    • Co-own the plan to turn an AI agent built for one setup into a reusable core that can scale across clients.
    ImpactGave the team its first measured quality baseline for the agent. Skipped tool calls are now partly fixed with engineering.
  2. Auriseg, an AI cybersecurity startup
    Sole PM
    Product Manager Intern, the only PM Mar – Jul 2026
    • Owned an AI security scanner for LLM and agentic apps end to end, from market research to public launch.
    • Built the phased product roadmap from two weeks of competitor and market-gap research.
    • Ran a free beta with existing clients, and set the launch pricing and packaging.
    • Rebuilt onboarding for non-security users: plain language, an in-product AI guide, and a real scan in the very first step.
    • Wrote PRDs and roadmaps, co-wrote functional specs with developers, and presented plans to senior leadership.
    ImpactLaunched publicly in 3 months, across 8 sprints and 2 milestones, as the company's only product manager.
Case studies

Work

Knit PM

Most PMs said it can't be solved. I built a demo to test that.

25 PMs and project managers interviewed

AI Assistant Audit

500 questions to a live AI assistant. 1 in 7 answers was made up.

500 questions asked and traced

Koo

It stopped paying for users, and lost two-thirds of them.

~10M peak monthly users

Quibi

A 90-day free trial hid the only number that mattered.

$1.75B raised before launch

Google Stadia

The streaming worked. The cost of trying a game didn't.

Nov 2019 launch

Gamma

Paying removes the badge that powers Gamma's growth.

$100M+ ARR

Fireflies.ai

Its biggest growth channel is people who never signed up.

20M+ users

Jar

The first failed auto-save decides whether the habit lives.

35M users

Etsy

7% of buyers drive 40% of sales. The warning comes late.

7% of buyers are habitual
Signal console

One signal per product

Click a row to open the case
signals.jetson
> scan --for silent-failure --products 9
ProductSignal I watchValueStatus
Knit PMDecisions reopened after sign-offIn pilotTesting
AI Assistant AuditTool-skip rate6.8%Measured
KooUsers after paid acquisition stopped-67%Public data
QuibiTime to first payment signal90 daysPublic data
Google StadiaCost to try one new gameFull pricePublic data
GammaBadge-off share of deck viewsTo testHypothesis
Fireflies.aiRecap pull vs. pushTo testHypothesis
Jar30-day survival after first failed saveTo testHypothesis
EtsyCadence slip of habitual buyers-9%Public data
How I work

Problem first. Always.

PM isn’t about jumping to solutions. It’s about asking more questions, the right ones. Spend 80% of your time on the problem, and the right solution becomes obvious.
Jetson Samuel, on how I work
  1. 01

    Ask

    Question the problem from the user, business and tech side before touching a solution.

    5 Whys · Jobs to be done
  2. 02

    Understand

    Spend most of the time here: interviews, data, and finding where the product quietly breaks.

    User interviews · SQL · funnels
  3. 03

    Pick the signal

    Choose the one metric that proves the problem is real, with a metric tree under it.

    North Star · metric trees
  4. 04

    Test small

    Run the smallest experiment or prototype that could prove me wrong.

    A/B tests · prototypes · evals
  5. 05

    Ship and learn

    Write the PRD, launch with clear success criteria, then review what the data says.

    PRD · launch review
Skills and tools

What I work with

Where I go deepest: making AI agents trustworthy enough to ship.

  • AI evals: golden datasets, rubrics, LLM-as-judgeStrongest
  • ADLC (agent development lifecycle)
  • Agentic workflows and tool calling
  • RAG design: retrieval, chunking, grounding
  • Prompt and context engineering
  • Model selection: quality, cost, latency
  • Guardrails and AI safety
  • Human-in-the-loop design
  • AI product metrics: groundedness, task success
  • Hallucination root-cause analysis
  • Fine-tuning vs. RAG decisions
  • Responsible AI and data privacy
Beyond work

Range

All achievements →
HICAS Music Band on stage, 25 members with instrumentsHICAS Music Band, 2022 to 20259+

First-place wins leading a 25-member band in inter-college competitions across Coimbatore.

Multiple Dots by JetsonMultiple Dots, 2025 to now33K

Followers in about a year across Instagram and YouTube, with paid campaigns for Scaler, Simplilearn and TaskTracker.

Thumbnail of the video People Who Dream Big But Do Nothing, in TamilBest video100K+

Views on “People Who Dream Big But Do Nothing”, without paid promotion.

Hiring a PM?

Let's connect the dots.

Read one case study. If it changes how you see that product, let's talk.

jetsonarmstrong@gmail.com
LinkedInResume (PDF)