AI, Data, and Optimization for Better Operations
Alpha Intelligence helps manufacturers, energy companies, and data-driven teams turn complex data into decisions, automation, and measurable improvement.
Industries: Manufacturing · Energy · Industrial Processes · ERP & Business Systems
AI agents for document intelligence, knowledge assistance, ERP support, and automation workflows. We build voice pipelines (TTS, STT, voice-to-voice), LLM evaluation and observability, token and cost optimisation, security hardening, and MCP integrations with LangGraph and LangChain.
Reliable forecasts, anomaly detection, and decision support from operational and sensor data. We apply deep learning, PINNs, computer vision, reinforcement learning, system identification, and hybrid models — with sensor calibration, ETL, feature engineering, and MLOps.
Mathematical optimisation, system identification, and Model Predictive Control for production, energy, and scheduling decisions. Applied across brick, glass, plastic-film, and carbon-fibre manufacturing, virtual-power-plant planning, and MILP-based production scheduling.
ETL from industrial sources including SAP HANA, OPC UA historians, and sensor networks. We design and deploy Azure-based pipelines, data validation, schema versioning, and Power BI reporting — making your analytics and AI production-ready.
Selected Projects
Agentic ERP Assistant — Voice-enabled AI assistant for ERP navigation, document Q&A, and controlled business process automation. Built on LangGraph, MCP, multi-provider LLMs (Claude, OpenAI, Ollama), Langfuse observability, PostgreSQL, Qdrant, and Azure. Manufacturing sector engagement over approximately one year.
AI Document Agents — Search, recommendation, and document refinement agents using RAG pipelines, OpenAI, Ollama, and Mistral. Delivered as rapid proof-of-concept projects for automotive and professional services use cases.
MLOps Smart Alerting — Production anomaly detection and drift analysis system with dynamic thresholds and automated model retraining pipelines. Deployed on Azure with continuous monitoring across multiple ML models in a live industrial environment.
Industrial MPC and Energy Forecasting — Model Predictive Control implementations for brick, glass, plastic-film, and carbon-fibre manufacturing processes. Virtual power plant scheduling and energy price forecasting for the energy sector.
Technology Stack
AI & Agents: LangGraph · LangChain · MCP · Langfuse · Claude · OpenAI · Ollama · Llama · Mistral
Data & Infrastructure: PostgreSQL · Qdrant · Redis · RabbitMQ · Docker · Azure · OPC UA · SAP HANA
ML & Science: Python · PyTorch · scikit-learn · Azure ML · Azure AI Foundry · Power BI
From Technical Challenge to Practical Delivery
How we work
We start with the business decision, data reality, and operational constraints—not a predefined tool.
Engagements can begin with a focused assessment, a prototype, or delivery into an existing client environment.
Typical outcomes include faster access to knowledge, more reliable forecasts, reduced process variability, and decision support that teams can use.