Product Builder & AI Product Engineer
Follow the
problem deeper.
I turn ambiguous problems into products,
systems, and working software.
The interesting work
usually lives between the boxes.
I don’t sit neatly
between the boxes.
That’s the point. I follow a problem through the layers it takes to solve it—from the person using the product to the systems underneath.
Model the work, not just the screen.
Follow the handoffs, decisions, exceptions, and feedback loops. Sometimes the right solution is a simpler workflow.
Working systems.
Real constraints.
Internal tools, operational workflows, and product systems. The useful question is what people needed—and what had to work.
Sales CRM
A working rhythm for leads, conversations, and follow-ups.
Album operations
Connecting the people behind a multi-stage production process.
Asset management
Turning a distributed physical operation into a trackable system.
Workflow & proposal product
From configurable workflows toward a broader media platform.
THE INTERESTING PART IS THE SYSTEM, NOT THE TECHNOLOGY TAGS.
Make it real.
See what happens.
I like turning ideas into systems. The process is deliberate; the path rarely stays perfectly straight.
What is really happening?
Start close to users, operations, and the places where existing software breaks.
How does the whole system work?
Map the workflow, its roles, its constraints, and the relationships between its parts.
What is the smallest useful change?
Choose the simplest intervention that addresses the actual problem. It may be code, automation, or a better workflow.
What happens if we make it real?
Prototype and implement. AI-assisted development is a multiplier for product and system reasoning.
Can someone actually use this?
Put the system into use. Deployment and operations are part of building the product.
What does reality say?
Pay attention to how the system works in practice and where its assumptions stop holding.
What should change next?
Use that feedback to improve the system. Return to understanding whenever the problem changes.
More layers.
More ownership.
My career didn’t move from one box to another. It accumulated layers. Earlier capabilities stayed useful as the problems became more complex.
- 01
Customer & support
ACCUMULATED - 02
Testing
ACCUMULATED - 03
Product
ACCUMULATED - 04
Internal systems
ACCUMULATED - 05
Development
ACCUMULATED - 06
Infrastructure
ACCUMULATED - 07
AI systems
EXPLORING
EARLIER CAPABILITIES STAY IN THE SYSTEM.
Open questions.
Working experiments.
Serious research, at different stages of development.
LAB / 01AI-assisted engineeringRepository harness
Persistent engineering structure for agents working across connected projects.
Building / Exploring
Repository harness
Persistent engineering structure for agents working across connected projects.
What if architectural context, standards, workflows, and reusable intelligence lived above individual repositories?
A central engineering control layer can encode project relationships, instructions, documentation, and testing expectations.
The goal is to improve how AI coding agents work across projects, rather than repeatedly rebuilding context.
CONCEPTUAL MODEL / EXPLORATION
- 01Context
- 02Projects
- 03Agents
- 04Feedback
LAB / 02Bounded decision-makingAutonomous systems
Studying agents where decisions have measurable consequences.
Experiment
Autonomous systems
Studying agents where decisions have measurable consequences.
How do you separate research from execution and keep autonomy inside explicit boundaries?
Research directions include market data, evaluation, risk controls, and model/API/local-model tradeoffs.
The initial trading experiment concept involved approximately ₹1,000. It is an agent-systems experiment, with no claim of investment expertise or profitability.
CONCEPTUAL MODEL / EXPLORATION
- 01Research
- 02Decision
- 03Risk boundary
- 04Execution
LAB / 03AI-native softwareMemory & agent systems
Intent, context, reasoning, tools, and actions as part of the product itself.
Research
Memory & agent systems
Intent, context, reasoning, tools, and actions as part of the product itself.
What does software become when it can remember context and act on intent?
Exploring agents, orchestration, MCP, memory architectures, and specialized decision systems.
Voice-first interaction and prediction-based interfaces are areas of interest, not finished capabilities.
CONCEPTUAL MODEL / EXPLORATION
- 01Intent
- 02Memory
- 03Tools
- 04Action
LAB / 04Hardware / livestockPhysical-world systems
Sensors, wearables, and the economics of systems beyond a screen.
Exploring
Physical-world systems
Sensors, wearables, and the economics of systems beyond a screen.
How does a useful device become an affordable, maintainable system?
Research includes virtual fencing, animal tracking, livestock wearables, monitoring, sensors, and multi-animal systems.
This is an area of exploration; no deployed hardware product is claimed.
CONCEPTUAL MODEL / EXPLORATION
- 01Animal
- 02Sensor
- 03Signal
- 04Decision
LAB / 05Operations / real-world businessTerral
Testing product thinking against supply chains and physical constraints.
Exploring
Terral
Testing product thinking against supply chains and physical constraints.
What changes when the system includes sourcing, processing, machinery, and markets?
An entrepreneurial exploration around food processing, agricultural products, dehydration, botanical products, and related operations.
Software is one kind of system. Manufacturing workflows, compliance, supply chains, and unit economics are others.
CONCEPTUAL MODEL / EXPLORATION
- 01Source
- 02Process
- 03Supply
- 04Market
Notes from
the workbench.
Thinking in public, when there is something useful to share.
Research notes
coming soon.
- AI-native software
- Agent harnesses
- Product engineering
- Internal tools & workflows
- Decision & memory systems
- Real-world automation
A few things
I come back to.
- 01
Build before over-planning
Reality gives better feedback than speculation.
- 02
Understand the workflow
Many product problems are workflow problems disguised as feature requests.
- 03
Technology is a tool
Use code where code makes sense. Use automation where automation is enough.
- 04
AI should change the product
A system that understands intent should change the shape of the experience.
- 05
Simple systems win
Especially when humans actually use them.
- 06
Stay close to reality
Production teaches things prototypes cannot.
I kept following
problems deeper
into the system.
I didn’t enter technology through one perfectly defined role. I started close to customers and support, where I saw how products actually fail in the real world.
That led me into testing, product, internal systems, development, infrastructure, and eventually AI. The progression was about owning more of the problem.
I enjoy understanding why something should exist, designing how it should work, figuring out how to build it, and seeing what reality says.
I like turning ideas into systems.
Work across the Huemn / ve AI ecosystem, spanning product, operations, internal tools, development, infrastructure, and AI-related systems.