Why This Matters
Every few months I audit my stack. Tools I've dropped, tools I've kept, tools I've added. In 2026 the tooling landscape moved fast — especially on the AI side — so here's where I actually am right now.
This isn't aspirational. Everything on this list is in active use across client work or my own products.
QA Automation Stack
This is the work that pays the bills. I test enterprise Java backends, REST APIs, and web frontends.
Core Frameworks
Java + TestNG — still the standard for enterprise QA. Verbose, but stable and easy to hand off to other engineers on a team.Selenium WebDriver — browser automation. I use it with Page Object Model, not raw driver calls.RestAssured — API testing in Java. Fluent DSL, great for building readable test chains.Infrastructure
Maven for build management and dependency resolutionGitHub Actions for CI — free for small projects, scales wellAllure Reports for test reporting. Stakeholders can actually read the results.What I Avoid
Cypress for serious test automation. Great DX, not suited for complex multi-service setups.Playwright gets close but the Java bindings are still immature compared to the Node.js version.
SaaS / Product Stack
For building TrackmyRival, Invoysr, Trovaly, and client projects.
Frontend
Next.js 14 (App Router) — the default choice. RSC, streaming, built-in API routes, Vercel deployment. Nothing else comes close for solo builders.TypeScript — non-negotiable. The overhead pays back immediately.Tailwind CSS — utility-first CSS that I can move fast with. No fighting with stylesheets.Framer Motion — animations without writing CSS keyframes.shadcn/ui — unstyled components I actually own. Not a component library I'm locked into.Backend / Data
Supabase — Postgres + auth + storage + realtime in one. The RLS policies take 30 minutes to learn and then it just works.Vercel — deploy target. Edge functions, environment variables, domain management. I don't think about infra.Payments & Comms
Stripe — credits-based billing for TrackmyRival, subscriptions for Invoysr. The API docs are the best in the industry.Resend — transactional email. 3000 free emails/month, React Email templates.
AI Engineering Stack
The newest part of the stack. Used in TrackmyRival, the Rovo Agent, and the WhatsApp Learning Bot.
LLM
Anthropic API (Claude) — my primary LLM for anything that requires reasoning, structured output, or following complex instructions. Claude 3.7 Sonnet is my default.I don't use OpenAI. The output quality difference matters for the types of tasks I build around.Data Extraction
Firecrawl — web scraping that returns clean markdown. TrackmyRival wouldn't exist without it. The managed API handles JS rendering, rate limiting, and proxies.Orchestration
Python + APScheduler for scheduled AI jobs (the WhatsApp bot)For most products I don't use LangChain or LlamaIndex — direct API calls with typed schemas are simpler to debug and maintain
Daily Dev Tools
WebStorm — still the best IDE for TypeScript/JavaScript. The refactoring tools are worth the subscription.Warp terminal — AI autocomplete in the shell is genuinely useful.Linear for personal project management — even solo, having structured issues helps.Loom for async client updates — faster than writing long emails.
What I'm Watching
Bun — the Node.js replacement. Not in production yet but runs my scripts locally.
Cursor — AI-assisted coding editor. I use it for boilerplate-heavy tasks.
v0 by Vercel — useful for scaffolding UI quickly, then I clean up the output.
Stack always evolving. The goal is fewer tools that do more, not a longer list.