01
Product surface
Interface
Complex web and mobile products, operational tooling and data-rich experiences built around real workflows.
- React
- Next.js
- TypeScript
- Vite
- React Native
- Expo
- Tailwind
- TanStack
- Zustand
- GSAP
I take complex problems,figure out what matters,and build the solution.
How I Work
I break complex problems down, understand how their parts relate, and focus on the area that needs deeper attention.I break complex problems down, understand their relationships, focus where deeper analysis is needed, shape the solution around the business goal, make it executable, then build, validate, and ship it.
How I Work
I break complex problems down, understand how their parts relate, and focus on the area that needs deeper attention.I break complex problems down, understand their relationships, focus where deeper analysis is needed, shape the solution around the business goal, make it executable, then build, validate, and ship it.
How I Work
From complexity to something executable.
Business goal
Understand the problem
Understand the relationships
Focus where needed
Shape the solution
Shape
Make executable
Build. Validate. Ship.
Selected Work
Different systems. Different constraints.
01
Commerce infrastructure
Breakers Store
Rebuilt a legacy gaming store into an automated commerce platform spanning payments, fulfillment, operations, and customer support.
150k+ monthly visitors
$3M+ monthly revenue
6× order volume
90%+ auto-fulfilled
View case study
02
Business operations
Vida
Turned an Excel-driven insurance brokerage into a connected operating system for policies, finance, teams, and operations.
Operations + finance unified
Manual workflows systemized
View case study
03
AI systems engineering
AI Ops Investigator
Built a production-oriented AI operations agent that investigates incidents autonomously while keeping authorization, execution, and verification deterministic.
228 tests passed
4/4 live safety scenarios
0 safety violations
Crash-safe automation
View case study
The NDA problem.
More than 90% of my professional work is covered by one.
The projects above are the systems I can discuss publicly — not the extent of the work I've done.
Technical Range
I work across the system.
Product → production
I move between interface, application logic, realtime systems, data and infrastructure depending on what the product actually needs.
02
Application logic
Systems
Business rules, APIs, authentication, integrations and realtime workflows behind the product surface.
- Node.js
- NestJS
- Fastify
- REST APIs
- GraphQL
- Apollo
- WebSocket APIs
- Prisma
- Zod
- JWT
03
State + realtime
Data
Reliable state across relational data, asynchronous communication, background processing, failure and recovery.
- PostgreSQL
- Redis
- MySQL
- MongoDB
- BigQuery
- RabbitMQ
- Pub/Sub
- Kafka
- Temporal
- Webhooks
04
Delivery
Infrastructure
The layer where the product becomes a production system that can be deployed, operated and evolved.
- GCP
- AWS
- Cloud Run
- Docker
- Kubernetes
- Terraform
- Cloudflare R2
- Nginx
- CI/CD
Across every layer
TestingValidationSecurityObservabilityAuditability
Applied AI
AI that can retrieve, reason, and act.
I treat the model as one component inside a larger product system — surrounded by context, retrieval, tools, validation and explicit application behavior.
01 / 06
AI product loop
01
Context
Understand what matters.
Start with user intent, product state and the information the model actually needs to make a useful decision.
User intent
Application state
Conversation context
02
Retrieve
Bring in the right knowledge.
Use retrieval when product knowledge, documents or domain context should ground the model instead of relying on its training alone.
RAG
Relevant documents
Domain knowledge
03
Reason
Let the model choose a path.
LLMs and agents interpret the available context, decide what is needed next and coordinate multi-step workflows.
LLMs
AI agents
Prompt design
04
Act
Give it useful tools.
Models become substantially more useful when they can interact with APIs, workflows and business capabilities rather than only generate text.
Tool calling
MCP
APIs
n8n
05
Constrain
Control what comes back.
Structured outputs, validation and explicit application boundaries keep generated behavior usable by the rest of the product.
Structured outputs
Schemas
Validation
06
Apply
Change the workflow.
The useful outcome is not the model response itself. It is the product action, document, decision or workflow the response enables.
Product action
Human review
Auditable output
Experience
OpenAI APILLM applicationsRAGAI agentsAI ScribeTool callingStructured outputsMCPPrompt engineeringn8n
The principle
The model is one component. The product is the whole loop.
About
I like the part where the problem is still messy.
Product engineer
My best work usually starts before there is a clean technical specification.
I work directly with founders, stakeholders and domain experts to understand the business, challenge assumptions and turn ambiguous requirements into something executable.
From there, I can stay hands-on across product, interface, backend, data, infrastructure and AI until the system is working in production.
Experience
6+ years building products across commerce, healthcare, fintech, government systems, business operations and 0→1 software — including technical leadership, architecture and end-to-end delivery.
01
Product
Understand the real problem before choosing the solution.
02
Engineering
Own enough of the system to make the right trade-offs.
03
Delivery
Stay close to the work until it is useful in production.
Contact
Have a difficult product problem?
If you're building something complex, replacing something that no longer works, or trying to turn an unclear idea into a real product, I'd be interested in hearing about it.
Good conversations usually start with
A messy problemA product decisionA system that needs rebuilding