I build products.My systems run them.
A product manager who ships AI systems that run autonomously. Three products live in production, trading real capital, tutoring real students, and carrying a decade of life context.
Autonomous Trading System
MARKET REGIME
LAST SCAN
POSITIONS HELD
UNIVERSE
LAST SIGNAL
// LAYER 5 - EXECUTION
Upstox adapter · Capital checks · Position limits · NSE order routing
// LAYER 4 - CLAUDE REASONING
Haiku (3× daily scans) · Sonnet (weekly review) · Trade decisions
// LAYER 3 - MEMORY
Position theses · Scan journals · Regime history · Trade log
// LAYER 2 - ANALYSIS (DETERMINISTIC)
Regime detection · Pre-screener · RSI/MACD/BB/EMA/ADX · Volume profile
// LAYER 1 - DATA
Live prices · Historical OHLCV · FII/DII flows · Economic calendar
// 01 - KAIROS
The Autonomous Trader
Real capital. No human at execution time.
Kairos runs 24/7 on NSE. Every few hours it scans a curated universe of stocks, performs multi-layered technical and macro analysis, and - if conditions are right - places a real BUY or SELL order with no human confirmation.
The system selects its own AI model: Claude Haiku for routine scans (speed, cost), Claude Sonnet for weekly portfolio reviews that require deeper reasoning. It remembers why it entered every position and reasons about exits with full context of its own history.
-
-
-
-
-
-
// 02 - MENTORIZON
The Life Context Engine
Not a chatbot. An intelligence that knows your whole life.
Most AI tools answer questions in isolation. Mentorizon maintains a persistent, cross-domain model of your life - career, health, finances, relationships, parenting - and routes your questions to specialist AI personas that advise you with awareness of the full picture.
Ask the Career specialist about a job offer and it knows about the financial context you shared with the Finance specialist last week. The advice is coherent because the context is shared.
User declares topic + desired outcome
LLM generates goal-anchored curriculum
sections → lessons
User completes section
LLM generates assignment
User submits answers
LLM evaluates + coaching feedback
Developing / Progressing / Strong
Mid-course: User gives direction feedback
LLM regenerates remaining curriculum
completed work preserved
// 03 - LEARNING0TO1
The Adaptive Curriculum
Rewrites itself based on how you actually learn.
Learning0to1 starts by asking what you want to be able to do - not just what you want to know. The curriculum is built around that outcome. Every section ends with an assignment; every submission triggers streaming feedback with per-dimension coaching, not just a score.
Mid-course, if you tell it to go deeper, move faster, or change direction - it regenerates everything you haven't done yet. What you've already completed stays intact.
Goal-anchored curriculum
Built around what you want to DO, not what you want to know.
Every section maps to your stated outcome.
Section-level assignments
Every section ends with a real assignment.
Streaming feedback with per-dimension coaching.
Mid-course adaptation
Regenerates remaining curriculum on your feedback.
Completed work stays intact.
// THE THREAD
Three products. One architecture.
Every system I build does the same thing: it holds context across time, reasons from memory, and acts autonomously when conditions are right.
Kairos remembers why it entered every trade and reasons about exits with that history. Mentorizon holds a persistent model of your whole life so each specialist advises you in context. Learning0to1 rewrites your curriculum based on how you actually responded - not what was planned.
The pattern isn't accidental. It's a belief: AI is only as useful as the memory it carries.
Kairos
MEMORY
Position theses, scan journals
AUTONOMOUS ACTION
BUY/SELL execution, no human loop
REAL STAKES
Real capital, real orders
Mentorizon
MEMORY
Persistent cross-domain life context
AUTONOMOUS ACTION
Specialist routing, context-aware responses
REAL STAKES
Honest life advice
Learning0to1
MEMORY
Curriculum history, learning response
AUTONOMOUS ACTION
Curriculum regeneration on feedback
REAL STAKES
Actual learning outcomes
// 04 - THE BUILDER
How I got here.
ACT 1 - THE FOUNDATION (2007–2015)
Java developer at L&T Infotech. Eight years implementing and customising Oracle Agile PLM for semiconductor and pharma clients across three countries. Learned that software only matters when it changes how people work.
ACT 2 - THE EXPANSION (2015–2022)
Joined Flex as a consultant, inherited a broken PLM system, rebuilt it, and rolled it out to 80+ global manufacturing sites. Then ran it - managing roadmap, 11-person team, competing priorities across a global stakeholder network. Crossed the line from technologist to product owner.
ACT 3 - THE BUILD (2022–PRESENT)
PM at Toppan Merrill leading the modernisation of SEC ownership reporting (SEC Connect → Quinn). 250+ clients, 20K+ annual filings, $2.5M ARR. Learned Figma independently when design resources were pulled. Outside work: built three AI systems from scratch - alone, in production, doing real things.
Relocating to Sydney · September 2026
Australian Permanent Resident (SC-189)
Full unrestricted work rights. No sponsorship required.
PRODUCT MANAGER
Capital Markets Reporting
Toppan Merrill · Oct 2023 – Present
- –SEC Ownership Reporting modernisation
- –SEC Connect → Quinn platform migration
CLIENTS
250+
ANNUAL FILINGS
20K+
ARR
$2.5M+
TEAM
8 eng · 5 QA
Self-taught Figma mid-delivery when design resources were pulled. Kept engineers unblocked through 7 of 10 forms.
Building something that needs a PM
who can also ship?
I'm relocating to Sydney in September 2026 and am available for conversations now.