Feed
Unified stream of what I'm building, writing, learning, and thinking about. Think of this as a living changelog for my work and interests.
Current Focus
- OpenCode AI coding agent with MCP server architecture
- Multi-agent orchestration framework for production workflows
- Personal AI knowledge base with RAG pipelines
- Multi-agent orchestration frameworks
- Advanced prompt engineering & LLM fine-tuning
- Kubernetes operators & custom controllers
- Building LLM Applications with Prompt Engineering
- Designing Machine Learning Systems (Chip Huyen)
- Reinforcement Learning: An Introduction
Activity Timeline
OpenCode AI Agent
Open-sourced OpenCode — an AI coding agent with MCP server architecture for tool-augmented codebase operations.
sushantdev.com v2 — Living Platform
Complete redesign: unified feed, editorial design system, case studies with Mermaid diagrams, content-driven current focus.
Building Production RAG Pipelines
Deep dive into hybrid search strategies, chunking techniques, and evaluation frameworks for RAG in production.
Multi-Agent Orchestration
Studying AutoGen, CrewAI, and LangGraph for building coordinated multi-agent systems.
Survey of Agent Frameworks
Notable paper: comprehensive survey of LLM agent frameworks comparing planning, memory, and tool-use capabilities.
Homelab AI Infrastructure
Documented full homelab setup: Proxmox, Ollama, vLLM inference server, monitoring stack, and automation.
Agentic AI: From Prototype to Production
Spoke at AI Eng Conference on patterns for taking AI agents from research demos to production systems handling real traffic.
Kubernetes Operators with Python
Learning to build custom Kubernetes operators using Kopf framework for AI workload orchestration.
Prompt Engineering Patterns for Production
Structured prompt techniques tested across thousands of production LLM calls: chain-of-thought, few-shot, routing, and guardrails.
Notes on Designing ML Systems
Working through Chip Huyen's ML Systems Design. Key insight: ML systems fail more from data/ops issues than model quality.
Why MCP Changes the Agent Game
Posted: Model Context Protocol gives AI agents a standardized way to discover and interact with tools. This is bigger than most realize.
LLM Fine-tuning Pipeline
Built automated fine-tuning pipeline with LoRA/QLoRA, evaluation benchmarks, and A/B deployment for domain-specific models.