Previous Full Stack Focus
I have consistently learned skills to create scalable web applications for the past decade. My focus hovered around Node.js, React, Next.js, and Typescript with an ultimate goal of building SAAS applications. As my expertise sharpened for frontend skills, I realized I unintentionally neglected to develop my backend knowledge. Because of this, my focus extended to learning about relational databases so that I could make whole full stack applications.
Stumbling into Distributed Systems
As I explored relational databases like MySQL and Postgresql, I stumbled across NoSQL databases. A fun fact about me is that I can never satisfy my curiousity, and I dove deep into understanding why NoSQL databases exists, when to use them, and how distributed systems were leveraged to add horizontal scalability to these databases.
For all intents and purposes, this new interest in distributed systems and databases was a side quest I never completed. To this day I love seeing the market of postgresql wire compatible horizontally scalable platforms like Neon, new OLAP datastores like Clickhouse, modern OTLP databases like TigerBeetle, and how all of these datastore software wrangles systems programming and distributed systems to offer new interfaces for developers.
Updating Interests
This background was to hopefully give more context to a shift I am doing in my career/professional focus. As AI agentic coding grows, most of my full stack skillset is being cannibalized with each new model release or harness update. Truthfully, My mind was burnt out on frontend work, but now that most frontend tasks can be "easily" accomplished by people without the decade of experience I developed, I am shook in the words of Mobb Deep.
To fight the looming AI shadow that makes my skillset less and less marketable, I am pivoting to embracing the existence of AI and focusing on learning how to build platforms that enable them. The North Star I am working towards is based on this mantra, if I can be a 1% developer in the world at developing platforms that aid Agents in doing work, I will be alright. For example, I built https://www.imaginate.run/ . It is a vibe coding application whose main value proposition is built on 3 things:
- a handrolled AI coding agent harness
- Inngest for dynamic workflow orchestration
- E2b for a sandbox for the agent to code in and host its code My new mantra pushed me into deepening my Go expertise, learning about systems programming, containers, and sandboxing with the purpose of building environments for agents to do work in. The first batch of work I am focusing on is sandbox platforms like E2b, Daytona, and Ona. OpenAI recently acquired Ona, Cursor is moving towards focusing on cloud agentic platform development so you can always make progress without work being bound to your local computer, and Daytona and E2b seem to have bright futures as more and more work is pushed into agents working in the cloud.
Background Agent Platform Development Progress
To become the Kobe "Bean" Bryant of background agent platforms, I have recently read "How Linux Works", "Systems Programming Essential in Go", and "Concurrent Programming in Go" to brush up on the Go skills I feel I will need. Now, I just started "Building an Orchestrator in Go from Scratch". The goal jumping out to me from this new set of knowledge will be to remix the project made by this book to supporting agentic coding workloads with the ultimate goal of my very first agentic platform that can be used to asynchronously create pull requests through cloud workloads.
Next Step
After conquering my first agent platform, I plan on learning about MicroVMs like Firecracker and how I an extend or create a new platform that leverages the vm isolation you gain from this software as opposed to the container based platform I am learning about in "Building an Orchestrator in Go from Scratch" book .
As I create this new software, I hope to document my journey. I figured worst case I will develop my Go, Kubernetes, and backend skills in a way that increases my work value, and best case I may be able to create a novel value proposition I can turn into a company. It would be awesome to develop a platform that becomes good enough for me to build the platform with in a recursive fashion.