Rajan Takiar
AI Product Manager
Agentic AI · AdTech · Platforms
I build products at the boundary between what a model can do and what a business can risk.
10+ years across advertising, SaaS and AI—from measuring enterprise campaigns to building agentic products and shipping three AI apps independently.
10+ yrs
Product & digital advertising
8M+
Users on a platform I managed
$50M+
Quarterly ad budgets measured
3 apps
AI apps shipped solo
Career path · newest first
How I got here
From building agentic products back through product, platforms, performance, and measurement.
Interview me without me
Six agents · one orchestratorAsk a screening question. Each agent holds a different part of my record, and if none of them has the answer it says so instead of inventing one.
Supabase Edge Function
{{ meshStatus }}
Ask it something it shouldn't know
System ready
Type a question, pick a suggestion, or click an agent on the board.
{{ m.who }}
{{ m.text }}
{{ m.cited }}
Selected work
Click to open a caseAction
Built the 0→1 product, prompts and agent workflows; wrote specs and product code alongside engineering.
Decision
Gated irreversible actions by consequence: reversible work runs automatically; committed budget needs approval.
Impact
Strengthened prospect pitches and generated multiple requests for seven-day B2B preview trials.
0→1Agentic platform built
7 daysRequested B2B preview trial
Agentic workflows
Prompt engineering
Human-in-the-loop
Claude Code
Attribution
DSP integrations
Action
Launched one privacy product from concept; separately owned pricing, packaging and funnel work on the scaled platform.
Metric
Managed a product serving 8M+ users and $24M ARR, using continuous tests to guide decisions.
Impact
Drove 50% revenue growth and improved retention by 15%.
8M+Platform users
$24MAnnual recurring revenue
+50%Revenue growth
+15%Retention
0→1
Consumer SaaS
Retention
A/B testing
Pricing
Agile
Action
Designed, built, shipped and monetized DareUp, Date Magnet and First Date Questions independently.
Product craft
Combined language and vision models with subscriptions, offline paths and user-controlled content boundaries.
Impact
Turned product ideas into working proof before asking an engineering team to spend a sprint.
3AI apps shipped solo
0→1From concept to store
React
TypeScript
Supabase
LLM APIs
RevenueCat
Play Store
See Date Magnet on Google Play
Action
Implemented brand-safety and verification systems for Fortune 500 advertisers and translated results for executives.
Metric
Measured more than $50M in quarterly budgets across ComScore, DoubleVerify and IAS.
Impact
Raised viewability from 40% to 60% and reduced wasted spend by 20–60%.
$50M+Quarterly budgets
40→60%Viewability
20–60%Waste reduced
ComScore
DoubleVerify
IAS
Viewability
Brand safety
Executive comms
Product proof: Date Magnet
Live on Google PlayA multimodal AI dating coach I took from idea to production. Three screens show the core loop.
01
Bring the contextUpload the real profile instead of describing it.
02
See the diagnosisVision analysis scores the profile and ranks what to fix.
03
Act with contextGenerate an opener grounded in a specific match.
How I structure an AI agent
The model is one step in a product system. Guardrails decide what it may do; irreversible actions return to the user.
01 / INPUTUser intentStructured request enters
02 / POLICYGuardrail checkCost of error evaluated
03 / AGENTTool executionReversible work proceeds
04 / GATEHuman decisionIrreversible work pauses
05 / OUTPUTResult + traceEvery action remains visible
Where do you draw the line on AI autonomy?
AI product decision demoSet a safety boundary for a real AI product, then watch that decision control what the agent can do.
DM
Shipped productDate Magnet · AI dating coach
Users upload dating profiles or conversations. The AI can analyze, rank and draft—but some actions affect real people, saved work or money.
Live on Google Play
Product questionWhat should the AI do alone—and when should a person take control?
01 Set the boundary first
Choose how much Date Magnet may do before asking the user.
Ask every timeAct alone
AI can do{{ canSummary }}
Must ask{{ askSummary }}
{{ verdict }}
02 Run a Date Magnet job
See the full decision policy
{{ a.n }}{{ a.u }}
03 Watch the decision trace
No job running yetSet the boundary, then choose a Date Magnet job. The trace will show what runs automatically and what stops for approval.
{{ taskTitle }}Live policy trace
{{ s.no }}
{{ s.t }}{{ s.cost }}
{{ s.status }}
04 Decision
{{ decisionTitle }}
{{ decisionBody }}
My rule
Let AI handle work you can undo. Pause before it affects another person, spends money, or changes saved work.
Core Competencies & Focus Areas
Where I create leverageProduct execution
0→1 strategy, roadmaps, requirements, prioritization and release decisions.
AI architecture
Agent workflows, prompt and context design, output quality and production evaluation.
Guardrail systems
Human-in-the-loop boundaries based on reversibility, cost and user trust.
AdTech & measurement
Programmatic systems, attribution, verification, publisher operations and campaign economics.
Hands-on prototyping
React, TypeScript, product code and fast prototypes that make scope tangible.
Focused scope
Product systems around models, rather than model research. BI and data-science collaboration, rather than daily analytics engineering.
Get in touch
Based in Montreal and working across North America. Feel free to reach out.
Rajan Takiar — Product · Agentic AI · Advertising
Montreal → North America