Saddam Saku
I build AI-powered products, ship full-stack SaaS, and explore the edges of what's possible with modern software.
— About
Building at the edge of AI & software
I'm a software engineer obsessed with the intersection of AI and product. I spend my time building tools that developers and builders actually use — from LLM-powered apps to open source libraries and autonomous agents.
My current obsession is the AI infrastructure layer — how agents reason, how tools get integrated, and how indie developers can use these primitives to build genuinely useful products without a VC war chest.
Currently Exploring
- Model Context Protocol (MCP)
- Multi-agent orchestration
- Edge inference with WASM
- Go for backend services & CLIs
- AI-native SaaS patterns
— Stack
Tools & Technologies
A curated set of technologies I use to go from idea to shipped product — covering AI/ML, full-stack web, infrastructure, and developer tooling.
— Projects
Things I've Built
A selection of projects across AI tooling, full-stack SaaS, and open source.
FastAgent
Open source Python framework for building autonomous AI agents with tool use, memory, and multi-step reasoning. Plugs into any LLM backend via a unified interface.
PromptOps
CLI tool for versioning, testing, and deploying prompts as code. Integrates with CI/CD pipelines so you can treat prompts like software.
VectorStore CLI
Command-line interface for managing embeddings across Pinecone, Qdrant, and Weaviate. Supports bulk upsert, semantic search, and namespace management.
RAG Benchmark
Open dataset and evaluation harness for benchmarking Retrieval-Augmented Generation pipelines across different chunking strategies and embedding models.
— Now
What I'm Building Now
A live “/now” page — inspired by nownownow.com
Building an AI Agent Framework
Designing a lightweight Python framework for multi-step agent workflows with pluggable tool registries and memory backends.
Exploring Model Context Protocol
Deep-diving into Anthropic's MCP spec to understand how to build interoperable tool surfaces for LLM agents.
Shipping a SaaS MVP in 30 days
Applying the 'boring startup' approach — pick a niche problem, validate fast, ship an MVP, then iterate based on real usage.
Learning Go for Backend & CLI Tools
Working through the Go standard library and building small services and CLIs. The simplicity, performance, and concurrency model are a great fit for backend tooling.
— Contact
Let's build something interesting together
I'm open to collaborating on AI projects, full-stack products, and open source work. If you have an idea, a problem to solve, or just want to talk tech — my inbox is open.
me@saddam.engineer