University of Jyväskylä, Finland

Bridging the gap between AI Research and Production Engineering.

I am an AI Systems Architect specializing in building production-grade Generative AI platforms. With a background in High-Performance Computing (HPC) and Simulation, I don't just wrap APIs - I build secure, cost-optimized, and observable AI infrastructure.

I focus on "Day 2" Operations: Governance, FinOps (Cost Modeling), Latency Optimization, and Role-Based Security.

My repositories are Reference Architectures, not just demos. Each includes C4 Models, ADRs (Decision Records), and Cost Analysis.

• DocuMind-Enterprise | Agentic RAG Platform

  • Architecture: Asynchronous FastAPI + LangGraph State Machine.
  • Key Pattern: Identity-Aware Retrieval. Implements "Citation-First" governance to eliminate hallucinations in regulated industries.

• Meridian | Context Orchestration Engine

  • Architecture: 4-Layer Cognitive Chain (Identity $\to$ Intent $\to$ Knowledge $\to$ Generation).
  • Key Pattern: Row-Level Security (RLS) for RAG. Filters vector retrieval based on user roles (CEO vs. CTO) to prevent data leaks.

• Swarm-Factory | Cloud-Native IIoT Platform

  • Architecture: Event-Driven Architecture (EDA) on Azure (Event Hubs + Serverless Functions).
  • Key Pattern: Spec-Driven Development. Uses OpenAPI/AsyncAPI contracts to decouple high-velocity telemetry from dashboard visualization.

• Poseidon-Link | Polyglot Marine Control System

  • Architecture: C++ (Physics) + Go (Telemetry) + React (UI).
  • Key Pattern: Safety-Critical AI. Uses a Go broker to sanitize AI voice commands before they reach the C++ physics engine.

• Hyperion | Physics-Guided Sales AI

  • Architecture: Hybrid Intelligence (Vectorized Pandas + GPT-4o).
  • Key Pattern: Zero-Hallucination. Anchors Generative AI sales pitches to a deterministic physics simulation engine.

• Pipeline-X | Hybrid Cloud Data Platform

  • Architecture: Apache Spark (Big Data) + Airflow + Qdrant (Vector Search).
  • Key Pattern: Unified Pipeline. Orchestrates dual-stream processing to keep SQL Analytics and Vector Embeddings in sync.

I believe in Architecture over Implementation details. My work is guided by:

• The "Zero-Legacy" Mindset: Every project includes ADR (Decision Records) so future teams understand why choices were made.
• FinOps First: AI is expensive. I implement token caching, routing, and quantization strategies to minimize OpEx.
• Visual Communication: I use C4 Models to translate code complexity for C-Level stakeholders.

• AI & Orchestration: Python, LangChain, LangGraph, Semantic Kernel, OpenAI, Ollama.
• Vector Infrastructure: Qdrant, Supabase (pgvector), Milvus.
• Backend & Systems: FastAPI (Python), Go (Golang), C++17, Apache Spark.
• Cloud & DevOps: Azure (Entra ID, Container Apps), Docker, Terraform (IaC), GitHub Actions.
• Frontend: React, TypeScript, TailwindCSS, Streamlit.

• LinkedIn: linkedin.com/in/nibir-1
• Email: nahasat.nibir@gmail.com