The tech, toys, and tools that keep me creating.

My go-to for modern, modular backends. NestJS gives a clean architecture, TypeScript safety, and quick prototyping. It made Seldon scalable while integrating multiple legacy systems.

Powerful API design tool. I like it because it gives clients exactly the data they need with less over-fetching, making frontends faster and easier to maintain. It also enables reactive programming by design, keeping data flows smooth and real-time friendly.

Rock-solid framework I used extensively for scalable APIs. It provides excellent performance and reliability, which was critical in projects like Nova where enterprise clients required trust and speed.

Multi-model database that handles graphs, documents, and key-values in one system. Gave me an edge in projects involving semantic search and AI-driven relationships.

The backbone for full-text and semantic queries. Perfect for AI-based search features, like matching job requests to profiles in Nova. I also used OpenSearch in Seldon, where it was a more cost-effective choice without sacrificing capabilities.

A lightning-fast columnar database optimized for analytics at scale. I like it for handling large volumes of structured and semi-structured data with real-time queries. It’s a strong fit when AI agents or dashboards require instant insights.

I use n8n to quickly sketch, test, and prototype data pipelines. It’s perfect for experimenting with workflows before moving them into production.

Simple, ultra-fast cache and messaging layer. It made microservices communication smooth and efficient. I also use it with GraphQL for reactive programming.

I use MLflow to track, manage, and version machine learning experiments. It helps me keep workflows organized and reproducible across projects.

A vector database purpose-built for AI applications. I like it because it handles embeddings at scale with blazing performance. It’s a natural choice for RAG pipelines and semantic search, making AI agents smarter and faster.

My preferred framework for building RAG pipelines and agents. It abstracts complexity while giving flexibility to orchestrate models, tools, and retrieval systems.

A fine-tuning approach that makes training large models cheaper and faster by using quantization. I use it to adapt LLMs to domain-specific contexts without the heavy infrastructure cost of full training.

A great solution for running LLMs locally or on-premise. I use it when security and data privacy are critical, allowing me to deploy language models without sending sensitive data to external APIs. Perfect for industries with strict compliance needs.

My favorite web framework. Combines React with SSR and API routes, which accelerates development. I used it in Nova and Seldon for modern and fast apps.

Best cross-platform toolkit for building mobile apps with great UX. It allowed me to build Lumina for iOS and Android with a single codebase, pixel-perfect and smooth.

I use TypeScript to write safer, more maintainable JavaScript. Its type system helps catch issues early and makes complex projects easier to manage.

My favorite utility-first CSS framework. It lets me design fast, consistent, and responsive interfaces without leaving the code. Combined with Next.js, it gives projects a clean, modern aesthetic while staying highly customizable.

I containerize everything to make deployments portable and predictable. Essential for all my SaaS and AI work.

When scaling matters, Kubernetes keeps services resilient and orchestrated. Great for projects with multiple microservices like Seldon.

When I need to run heavier containers that aren’t always on, Fly.io is my go-to. It spins them up globally only when needed, making it easy to stay efficient without paying for idle resources.

For lightweight, production-ready pods, Koyeb handles them smoothly and reliably, making deployment simple and fast.