I'm an Agentic AI Developer at Panaversity building autonomous AI systems that take action, not just respond. I have completed all 6 progressive hackathons — evolving a simple file watcher into a full Kubernetes-orchestrated platform with Constitutional AI safety, Apache Kafka event streaming, Dapr service mesh, and a Discord bot.
Now I'm entering the next phase: building Digital Agent Factories — turning AI protocols (MCP, A2A, Agent SDKs) into production-ready digital employees (FTEs) using spec-driven automation.
Hackathons Completed : 6/6 ✅ (Bronze → Platinum)
Kubernetes Services : 14 running in 6GB cluster
Tests Passing : 180+
Architecture : Event-Driven + Constitutional AI
Methodology : Specification-First Development
Current Phase : Agent Factory & Digital FTEs |
Currently Building:
|
In the AI era, the most valuable companies won't sell software — they'll manufacture AI employees, powered by agents, specs, skills, MCP, autonomy and cloud-native technologies.
The shift from developer-as-typist to developer-as-orchestrator is here. I'm building systems where natural-language specs drive autonomous agents that don't just respond — they act, coordinate, and deliver.
| Concept | Description |
|---|---|
| Digital FTEs | AI agents that function as full-time digital employees, handling end-to-end workflows |
| Agent Skills | Reusable, composable capabilities (39 skills across 8 categories on Claude.ai) |
| Spec-Driven Automation | From manual coding to specification-first development — the spec IS the product |
| Agent Protocols | MCP (Model Context Protocol) & A2A (Agent-to-Agent) for standardized agent communication |
📊 Reference: Agent Factory: Building Digital FTEs — Presentation
A production-grade, Kubernetes-orchestrated AI application built progressively across all 6 hackathons (now complete), featuring Constitutional AI safety, event-driven microservices, and multi-interface access.
graph TD
subgraph "User Interfaces"
A["Discord Bot<br/>(TodoMaster AI)"]
C["Next.js Frontend"]
end
subgraph "Core Platform"
B["FastAPI Backend<br/>(+ Dapr Sidecar)"]
I["Constitutional AI<br/>Middleware"]
end
subgraph "Data Layer"
D["PostgreSQL 15<br/>(StatefulSet)"]
H["Redis 7<br/>(State Store)"]
end
subgraph "Event Streaming"
E["Apache Kafka<br/>(Strimzi KRaft)"]
F["Notification<br/>Service"]
end
subgraph "Observability"
G["Prometheus"]
end
A -->|REST API| B
C -->|REST API| B
I -->|Block/Flag/Allow| B
B -->|SQL| D
B -->|State| H
B -->|Dapr Pub/Sub| E
E --> F
G -->|Scrape| B
G -->|Scrape| F
style A fill:#5865F2,color:#fff
style C fill:#000000,color:#fff
style B fill:#009688,color:#fff
style I fill:#ff6b6b,color:#fff
style D fill:#336791,color:#fff
style H fill:#DC382D,color:#fff
style E fill:#231f20,color:#fff
style F fill:#4ECDC4,color:#fff
style G fill:#e6522c,color:#fff
Key Differentiators
| Feature | Implementation |
|---|---|
| Constitutional AI | Blocks homework-solving queries with Socratic responses, flags edge cases for human review |
| Zero-Code Infra Swap | Switched pub/sub from Redis → Kafka by changing 1 YAML file (Dapr abstraction) |
| 14 Services in 6GB | Full production stack at 44% memory utilization on Minikube |
| Event-Driven Audit | Every interaction published to Kafka with 24h retention |
| Multi-Interface | Same backend serves Next.js frontend + Discord bot (TodoMaster AI) |
Hackathon Progression — All Complete ✅ (Bronze → Platinum)
| Hackathon | Project | Tier | What I Built | Tests |
|---|---|---|---|---|
| H0 ✅ | Personal AI CTO | Bronze |
File watcher, auto-categorization, HITL approvals | 7/7 |
| H1 ✅ | Course Companion | Silver |
FastAPI backend, Constitutional AI filter, conversation tracking | - |
| H2 ✅ | AI-Powered Todo | Silver |
Spec-driven development, AI spec generation, CRUD with constitution | - |
| H3 ✅ | Advanced Todo | Gold |
Event-driven architecture, Kafka, Dapr, team collaboration | 149/149 |
| H4 ✅ | Cloud-Native | Platinum |
Full Kubernetes cluster (14 manifests), CI/CD, Prometheus | - |
| H4.5 ✅ | Discord Bot | Extended |
TodoMaster AI with 6 slash commands, K8s deployment | 31/31 |
| Project | Stack | Description |
|---|---|---|
| Physical AI Textbook Platform | Next.js, FastAPI, RAG, Gemini API | Interactive textbook with semantic search and context-aware RAG chatbot |
| LearnFlow AI Platform | Microservices, FastAPI, K8s, Docker | 5 specialized AI agents for personalized programming education |
| Course Companion FTE | FastAPI, ChatGPT API, Zero-Backend | Constitutional AI rules for LLM-based course management |
| Claude.ai Skills Marketplace | 39 Skills, 8 Categories | Reusable agent skills — document processing, automation, dev tools |
| CMT Stitching System | TypeScript, Next.js | CMT Stitching & Packing Management System |
| RepoToVideo | Python, AI | Turn any GitHub repository into a viral promo video with AI |
| Mathematics for AI | Python, Educational | Comprehensive repository covering mathematical foundations of AI |
Full Stack Breakdown
const stack = {
languages: ["Python", "TypeScript", "JavaScript"],
frontend: ["Next.js 14", "React", "Tailwind CSS"],
backend: ["FastAPI", "Node.js", "Uvicorn"],
ai: ["Constitutional AI", "RAG Systems", "LangChain", "LangGraph", "MCP", "A2A"],
agentSDKs: ["Claude Agent SDK", "OpenAI Agents SDK", "Google ADK"],
agentFrameworks: ["LangGraph", "CrewAI", "AutoGen", "OpenAI Swarm"],
databases: ["PostgreSQL 15", "Redis 7", "Vector DBs (Pinecone, Chroma, Qdrant, Weaviate)"],
infrastructure: ["Kubernetes", "Docker", "Dapr", "Helm"],
streaming: ["Apache Kafka (Strimzi KRaft)"],
monitoring: ["Prometheus", "Grafana", "OpenTelemetry"],
cicd: ["GitHub Actions (test → build → validate → security)"],
bots: ["discord.py (slash commands)"],
apis: ["OpenAI", "Claude (Anthropic)", "Google Gemini"],
protocols: ["MCP (Model Context Protocol)", "A2A (Agent-to-Agent)", "REST", "Dapr Pub/Sub"],
architecture: ["Microservices", "Event-Driven", "API-First", "Infrastructure-Agnostic"],
methodology: "Specification-First Development",
nextPhase: "Digital Agent Factory Builder"
};mindmap
root((2026 Focus))
Agent Factory & FTEs
Digital Employees
Spec-Driven Agents
Agent Skills Marketplace
Monetizing AI Knowledge
Agent Protocols
MCP
A2A Protocol
Claude Agent SDK
OpenAI Agents SDK
Google ADK
Multi-Agent Systems
LangGraph
CrewAI
AutoGen
OpenAI Swarm
Observability
OpenTelemetry
Grafana Stack
Loki + Tempo
Vector Databases
Pinecone
Qdrant
Chroma
Weaviate
Edge AI
WebAssembly
ONNX Runtime
On-device LLMs
Platform Engineering
Backstage
Crossplane
Terraform
| Area | Technologies | Why It Matters |
|---|---|---|
| 🏭 Agent Factory | Digital FTEs, Agent Skills, Spec-Driven Automation | Building AI employees that handle end-to-end workflows autonomously |
| Agent Protocols | MCP, A2A (Google/Linux Foundation), Claude Agent SDK, OpenAI Agents SDK | Standardizing how AI agents communicate and use tools |
| Multi-Agent Systems | LangGraph, CrewAI, AutoGen, OpenAI Swarm | Orchestrating specialized agents for complex workflows |
| Observability | OpenTelemetry, Grafana Stack (Loki + Tempo) | Unified telemetry for AI-native applications |
| Vector Databases | Pinecone, Qdrant, Chroma, Weaviate | Scaling RAG systems to production |
| Edge AI | WebAssembly (Wasm), ONNX Runtime | Running inference at the edge without cloud dependency |
| Platform Engineering | Backstage, Crossplane, Terraform | Building internal developer platforms for AI workloads |
| AI Safety | Constitutional AI, RLHF, Human-in-the-Loop | Ensuring AI systems are safe and aligned |
- Complete all 6 Panaversity Hackathons (Bronze → Platinum) ✅
- Build cloud-native system with Kubernetes, Kafka, and Dapr ✅
- Implement Constitutional AI safety with Human-in-the-Loop ✅
- Build event-driven architecture with Apache Kafka & Dapr ✅
- Deploy 14 services in Kubernetes with CI/CD pipeline ✅
- Build Discord bot (TodoMaster AI) with K8s deployment ✅
- Build Digital Agent Factory with MCP, A2A & Agent SDKs
- Build multi-agent system with MCP and A2A protocols
- Contribute to 3+ open-source AI/ML projects
- Publish 24+ technical articles and videos
- Launch course on Specification-Driven AI Development
- Grow YouTube channel to 1K+ subscribers
- Build and ship production Digital FTEs (AI employees)
"Traditional approach: Avoid AI mistakes. My approach: Learn FROM AI mistakes. Because real innovation happens at the edges of failure."
"We don't just teach people how to code; we are teaching them how to build and monetize Digital Agent Factories."
| Principle | Practice |
|---|---|
| Spec-First | No code without a specification |
| Production Quality | Every project is deployment-ready |
| AI as Collaborator | Not just a tool — a thinking partner |
| Open Source | Share knowledge, elevate the community |
| Agent Factory Mindset | From manual coding to spec-driven automation |
Topics: Agentic AI | Agent Factory & Digital FTEs | Spec-Driven Development | Cloud-Native Architecture | Constitutional AI Safety | Multi-Agent Systems | MCP & A2A Protocols
I'm open to collaborating on AI/ML projects, cloud-native systems, Agent Factory development, and hackathon partnerships.