From ambiguityto production.

I’m a senior full-stack software engineer with a product mindset. I work across frontend, backend, infrastructure, data, and AI, connecting technical decisions to real problems and owning the work through production.

  • ProductScoping, MVPs, UX thinking
  • EngineeringFull-stack, distributed systems
  • DataPipelines, Postgres, Kafka
  • AILLM apps, RAG, agents

Problems cut across layers. So do I.

  1. 1/4Shaping the product

    Product

    • Ambiguous problems
    • Stakeholder alignment
    • Product scoping
    • MVP definition
    • Prototyping
    • UX thinking
  2. 2/4Designing & shipping

    Engineering

    • System design
    • Data modeling
    • APIs & SDKs
    • Full-stack delivery
    • RAG & agent workflows
    • Production ownership
  3. 3/4Building the tools that build

    Platform

    • Internal platforms
    • Workflow automation
    • Agent infrastructure
    • Validation systems
    • Developer experience
  4. 4/4Running lean & secure

    Operations

    • Cloud cost
    • Performance
    • Observability
    • Reliability
    • Hardening
    • Access & secrets

What I build with.

  • TypeScript
  • Node.js
  • PostgreSQL
  • Microservices
  • pytest
  • Linux
  • Prometheus
  • LLM APIs
  • Python
  • React
  • MySQL
  • Event-driven
  • Jest
  • AWS
  • Grafana
  • RAG
  • Go
  • Next.js
  • Redis
  • Distributed systems
  • go test
  • Docker
  • CloudWatch
  • LangGraph
  • PHP
  • FastAPI
  • Kafka
  • Background workers
  • Playwright
  • Kubernetes
  • Observability
  • LangSmith
  • Bash
  • Prisma
  • Polars
  • SDKs
  • Selenium
  • Terraform
  • Prompt engineering
  • Data pipelines
  • Cloudflare
  • Tool calling
  • Query optimization
  • CI/CD
  • Structured outputs
  • Embeddings
  • Code agents
  • MCP

5 tools in Languages.

Selected work.

Products, platforms, and the engineering behind them.

  1. 01Travel tech, 0 to 1

    Plan Patagonia

    Product & Technical Founder

    A live travel platform bringing destinations, trails, stays, experiences, and practical travel tools into one Patagonia planning experience.

    • TypeScript
    • Next.js
    • Prisma
    • Neon Postgres
    • Claude API
    Plan Patagonia project preview
    Plan PatagoniaLaunched Aug 2026
    What I built

    I led product strategy, experience design, and engineering, taking Plan Patagonia from concept to launch.

  2. 02Engineering platform

    Workflow Execution Engine

    Open source, thesis

    From engineering knowledge to executable workflows. WEE runs versioned, replayable AI pipelines with explicit budgets. Already used to review and open real pull requests, end to end.

    • Go
    • TypeScript
    • JSON Schema
    • LLM APIs
    Workflow Execution Engine project preview
    Workflow Execution EngineLast commit Oct 26
    Behind the build

    core/tool/git deliberately has no push op. Landing a fix on GitHub goes through the generic http tool speaking GitHub’s own Git Data API, instead of a one-off integration bolted onto git.

  3. 03Backend tooling

    go-kafka-sdk

    Open source

    Every Kafka service re-implements the same connection, retry and error handling, each with its own bugs. This SDK packages those patterns once, so services ship events instead of plumbing.

    • Go
    • Apache Kafka
    Your applicationGo service, producerGo service, consumerNewProducer()producer.Produce()NewConsumer()consumer.Consume()go-kafka-sdkProducer APIConsumer APIsends messagereceives messageKafka cluster
    go-kafka-sdkLast commit Nov 10
    Behind the build

    Defaults to acks=all and auto.offset.reset=earliest: safe out of the box, one config map away from full librdkafka control.

Generating options is easy now. Choosing is the job.

AI creates countless possibilities. I bring the judgment to pick one, then own it all the way to production, so nothing gets lost in a hand-off.

  1. 1/4

    AI makes options cheap. Ideas, drafts and prototypes everywhere.

  2. 2/4

    I clarify the problem and make the product calls.

  3. 3/4

    One path, built and shipped. I own it end to end.

  4. 4/4

    Done means a number moved.

I write to work things out.

On engineering, product, and the parts of the job I’m still figuring out.

  1. Stop using the best model for everything: designing an AI stack around cost, risk, and complexity

    Essay2026

    The strongest model has plenty of work to do. It just should not be responsible for everything.

  2. AI didn’t replace our intelligence. It challenged our self-importance.

    Essay2026

    Answers have become abundant. Abundance is not the same as meaning.