AJ — SAP Architect & AI Systems Builder

Agentic RAG Pipelines Transformer Inference at the Edge Multi-Agent Orchestration Vector Embedding Space · Qdrant S/4HANA Clean Core Architecture Context-Aware Prompt Routing Deterministic Execution Graphs LLM Observability · LangFuse SAP BTP AI Core · GenAI Hub Chain-of-Thought Orchestration Semantic Agent Handoff Protocol ABAP for SAP HANA · Certified Agentic RAG Pipelines Transformer Inference at the Edge Multi-Agent Orchestration Vector Embedding Space · Qdrant S/4HANA Clean Core Architecture Context-Aware Prompt Routing Deterministic Execution Graphs LLM Observability · LangFuse SAP BTP AI Core · GenAI Hub Chain-of-Thought Orchestration Semantic Agent Handoff Protocol ABAP for SAP HANA · Certified

About

Driving Digital &
Intelligent Enterprise Transformation.

I'm a SAP S/4HANA Technical Architect with 18+ years of enterprise experience — global S/4HANA rollouts, enterprise consulting, complex ABAP landscapes — and an obsession with making those systems genuinely intelligent.

On the enterprise side: Clean Core ABAP, S/4HANA, SAP RISE, BTP, Fiori, and SAP AI Core — SAP certified since 2009, HCP/IoT Hackathon winner, multiple delivery excellence awards. On the AI side: a self-hosted inference cluster running agentic RAG pipelines, transformer models at the edge, multi-modal agent meshes, and vector embedding spaces — all on personal hardware, all observable, all in production.

Most architects know SAP. Most AI engineers don't. I'm the intersection. That's the edge.

// Enterprise

ABAP S/4HANA SAP RISE Clean Core SAP BTP AI Core GenAI Hub Fiori

// AI / Infra

Python LLM Agents RAG / Qdrant LangFuse n8n Docker Terraform

aj@hub ~ system.status

❯ aj status --verbose

role S/4HANA Architect · BTP · AI

experience 18+ years enterprise SAP

focus SAP RISE · Clean Core · AI Core

background Enterprise Consulting

# Active AI Stack

inference Local · GPU-backed

models Multi-model · rotating pool

agents orchestrating ✓

memory Vector store · live

traces Observability · live

uptime Always-on

certs "ABAP HANA · Fiori · S/4"

status ACTIVE · CONSULTING

❯

Projects

Open the box.

Enterprise SAP, inference clusters, agent meshes — decades of depth across domains. Battle-tested, iterated in production, not assembled from a tutorial.

Complexity · Expert

AJ Hub — Autonomous AI Orchestration Runtime

Not a dashboard — a fully typed multi-agent coordination runtime built from scratch. Coordinates 6+ heterogeneous AI agents (Claude Code, Cursor, n8n, Ollama workers, shell executors) via a custom event-bus and structured handoff packet protocol. Each agent receives typed task manifests, executes over SSH, and emits structured completion signals back into the activity feed. Features stall detection, sub-task decomposition, real-time SSE streaming, and SQLite-backed task persistence. Runs continuously on self-hosted infrastructure with local GPU inference.

Node.js Event-Bus Architecture Local GPU Inference SSE Streaming 6+ Agent Types Always-on pipelines

Complexity · Advanced

Hermes + Odysseus Agent Mesh

Custom inter-agent communication protocol on Docker overlay networks. Hermes is a typed message broker; Odysseus is the agent execution framework. Agents subscribe to channels, execute stateful multi-step sessions, and emit typed completion events — solving agent coordination without third-party frameworks that couldn't handle stateful sessions.

Python Docker Overlay Message Protocol

Complexity · Advanced

Agentic Listing Evaluation Pipeline

5-stage agentic pipeline: ingestion → configurable qualification gates → LLM relevance scoring → human review checkpoint → Jira tracker. Surfaces high-confidence matches in a periodic digest. n8n orchestration, Ollama inference, structured observability at every step.

n8n · Ollama Qualification Gates Human Checkpoint 5 configurable gates

Complexity · Advanced

SAP BTP + AI Core — IaC Deployment

Full Terraform IaC for SAP AI Core on BTP: subaccount provisioning, role collections (sap.default origin binding), resource groups, GenAI Hub LLM deployments, HANA Cloud, and AI Launchpad. Diagnosed and resolved an undocumented role-origin binding bug that had no SAP documentation — required reverse-engineering BTP IAM behavior.

SAP BTP Terraform AI Core · GenAI Hub HANA Cloud

Complexity · Mid-Senior

Full-Stack LLM Observability Layer

Custom instrumentation wrapping every AI call across 6 services — Groq, Ollama, Claude, LiteLLM all emit structured traces with session IDs, token counts, latency breakdowns, and cost attribution. Includes a post-session reflect agent (reflect.js + Groq) that auto-extracts learnings from traces and persists to vector memory. Every pipeline run is fully auditable.

LangFuse Custom Instrumentation 6 LLM Providers

SAP Official · 2016 Winner

SAP HCP / IoT Hackathon — 1st Place

Won SAP's official HCP/IoT Hackathon in Bengaluru (2016) among 16 competing teams. Built a real-time IoT sensor pipeline on SAP HANA Cloud Platform — ingesting device telemetry, processing in-stream on HANA XS, and surfacing live insights on a Fiori UI5 dashboard. Pre-commercial tooling meant building most of the integration layer from scratch.

SAP HCP HANA XS SAP IoT SAPUI5 1st of 16 teams

Services

How I can help.

Strategic advisory at the intersection of SAP expertise and AI capability.

SAP Architecture Advisory

S/4HANA design, ABAP modernization, and BTP integration strategy for enterprises navigating complex SAP transformations.

// strategic advisory

AI Integration

Connecting LLMs, agents, and AI pipelines to enterprise systems. RAG, vector search, agentic workflows — built for production, not demos.

// full implementation

AI Infrastructure

Self-hosted AI stacks, observability layers, and agent orchestration. From architecture to running deployment — I've built it for myself, I can build it for you.

// infrastructure design

AJ

Driving Digital &Intelligent Enterprise Transformation.

Open the box.

How I can help.

Let's build something.

Driving Digital &
Intelligent Enterprise Transformation.