I'm Ranit Saha, better known online as Coderooz.
I'm a software developer interested in building practical, scalable and intelligent software systems. I enjoy taking an idea from architecture and experimentation through implementation, testing, automation and deployment.
My work spans several areas, including:
- 🤖 AI-assisted development & agent systems
- 🧠 AI workflows, orchestration & multi-agent systems
- 🔌 MCP servers & developer tooling
- 🌐 Full-stack web applications
- 📱 React Native / Expo applications
- 🐍 Python systems & data processing
- 🗄️ Databases, APIs & backend architecture
- 📊 Financial data, analytics & visualization
- 🧩 Reusable frameworks, libraries & developer infrastructure
- 🏗️ Software architecture, governance & automation
I am particularly interested in the space where software engineering, AI agents and automation converge.
My projects are not limited to individual applications. I am increasingly interested in building the infrastructure and systems that allow applications and AI agents to work together reliably.
Exploring architectures for AI agents that can:
Understand a goal
↓
Build context
↓
Plan
↓
Delegate
↓
Execute
↓
Validate
↓
Recover
↓
Learn
This includes experimentation with:
- Agent orchestration
- MCP-based tooling
- Persistent agent memory
- Project context systems
- Sub-agent coordination
- Automated validation
- AI development workflows
- Engineering governance
I also build tools intended to make software development more structured and reusable:
- MCP servers
- Developer automation
- Template systems
- Reusable UI/editor components
- Project architecture tooling
- Testing and validation systems
- Documentation and reference systems
- AI-agent development infrastructure
A selection of projects and systems I've worked on or am actively developing.
An experimental platform exploring AI model organization, intelligence layers, learning systems and agent-oriented architecture.
Focus areas include:
- Model visualization
- Learning integration
- Intelligence layers
- Agent workflows
- Research infrastructure
- System observability
A hyperlocal commerce platform designed around an offline-first approach.
The ecosystem explores:
- Local commerce
- Inventory management
- Sales & purchasing
- Customer workflows
- Credit management
- Offline-first mobile experiences
- Administrative tooling
A persistent MCP-based infrastructure project for AI-agent context, project state and multi-agent coordination.
The architecture explores:
- Persistent agent memory
- Project identity
- Agent registration
- Task coordination
- Activity tracking
- Resource locking
- Conflict detection
- MCP tooling
A reusable rich-text editing system built around Next.js and Tiptap, with multiple editor modes, extensions, toolbars and reusable UI components.
An experimental Python project exploring the encoding and extraction of digital data through audio frequencies.
The project investigates:
- Binary framing
- Audio modulation
- BFSK
- Demodulation
- Round-trip encoding
- Signal processing
- Protocol experimentation
Software Architecture
│
├── Full-Stack Systems
│
├── AI / Agent Systems
│ ├── MCP
│ ├── Agent Memory
│ ├── Orchestration
│ └── Multi-Agent Workflows
│
├── Developer Infrastructure
│ ├── Automation
│ ├── Templates
│ ├── Tooling
│ └── Validation
│
├── Data & Analytics
│
└── Experimental Computing
├── Audio/Data Encoding
├── AI Research
└── New Interfaces
I prefer building systems that are:
- Modular — clear boundaries and responsibilities
- Testable — behavior should be verifiable
- Observable — failures should be diagnosable
- Automated — repetitive work should be delegated
- Documented — architecture should remain understandable
- Composable — components should be reusable
- Maintainable — today's solution should not become tomorrow's bottleneck
A recurring principle in my projects is:
Build the system, not just the feature.
