Practical, technical writing on production AI: forecasting, agentic systems, RAG, MLOps, and the rest of what we work on.
Rules-based chatbots fail enterprise operations because they cannot reason over live data. An agentic architecture with data grounding, action execution, and proactive alerting delivers the reliability that operational teams require.
Read more: Designing Agentic AI for Enterprise Operations: Architecture, Grounding, and Trust →
How Combining Foundational and Non-Foundational Models Can Take Forecasting Further
Read more: FORGE: A Hybrid Framework Bridging Foundational and Non-Foundational Models for Forecasting →Most fashion brands settle for 60-70% forecast accuracy. Here's why that's costing you millions and how AI can push accuracy above 85%.
Read more: Why 70% Forecast Accuracy Isn't Good Enough for Fashion Retail →Beyond revenue and margins, these metrics reveal the health of your inventory management and where you're leaving money on the table.
Read more: 5 KPIs Every Fashion Brand Should Track (But Most Don't) →47 posts
Rules-based chatbots fail enterprise operations because they cannot reason over live data. An agentic architecture with data grounding, action execution, and proactive alerting delivers the reliability that operational teams require.
Read more: Designing Agentic AI for Enterprise Operations: Architecture, Grounding, and Trust →Traditional demand planning relies on spreadsheet-based statistical models, manual reconciliation, and reactive replanning. Horizon replaces this with ML-driven hierarchical forecasting, scenario simulation, and automated action recommendations. Here is exactly what changes.
Read more: Horizon vs. Traditional Demand Planning: A Technical Comparison →Horizon and Pliney are not independent products — they form a connected intelligence layer. Horizon produces forecasts and recommendations; Pliney makes them accessible through natural language. This post describes how the systems integrate and what the architecture looks like in production.
Read more: ZAAI System Architecture: How Horizon and Pliney Work Together →How Combining Foundational and Non-Foundational Models Can Take Forecasting Further
Read more: FORGE: A Hybrid Framework Bridging Foundational and Non-Foundational Models for Forecasting →Most fashion brands settle for 60-70% forecast accuracy. Here's why that's costing you millions and how AI can push accuracy above 85%.
Read more: Why 70% Forecast Accuracy Isn't Good Enough for Fashion Retail →Beyond revenue and margins, these metrics reveal the health of your inventory management and where you're leaving money on the table.
Read more: 5 KPIs Every Fashion Brand Should Track (But Most Don't) →Running a promotion without simulating the impact is like jumping without looking. Here's how leading brands test promotions before committing.
Read more: Stop Guessing: Use Scenario Simulation for Promotion Planning →Multi-level demand forecasting requires more than accurate models—it demands a reconciliation strategy that preserves coherence across SKU, customer, and aggregate levels while optimizing for business impact.
Read more: Hierarchical Forecasting: How We Reconcile SKU, Customer, and Aggregate Demand →Predictive models tell you what will happen; causal models tell you what will happen if you change something. For pricing optimization, that distinction is everything.
Read more: Causal ML for Pricing: Beyond Predictive Models →Single-agent LLM systems struggle with complex compliance workflows. A multi-agent architecture with document routing, specialized agents, and research validation delivers the reliability enterprises require.
Read more: Building Multi-Agent Systems for Enterprise Compliance →Building production computer vision systems requires more than good models—it demands a pipeline architecture that handles ingestion, training, and cross-cloud delivery at scale.
Read more: Computer Vision at Scale: From Detection to Game Awareness →Before we build a single model, we spend 2–6 weeks understanding your data. Here's what we look for, what we commonly find, and why skipping this phase guarantees garbage-in-garbage-out.
Read more: Why Every ZAAI Project Starts with Data Quality Assessment →Enterprise AI projects carry real risk. Here's how we structure engagements to validate before committing: research phases, phased contracts, MVP-first delivery, and gated milestones.
Read more: The De-Risking Playbook: Research Phases, Phased Contracts, and MVP-First Delivery →135,000 transactions per year. 7,000 invoices per month. 5 banks. The hidden cost isn't just FTE time—it's errors, late detection, and audit chaos. Here's the automation path.
Read more: The Real Cost of Manual Financial Reconciliation (and How to Automate It) →Most OSA initiatives fail because they're reactive. Stockouts are discovered in-store, root causes are unknown, and data is disconnected. The fix: prediction, classification, and proactive alerting.
Read more: On-Shelf Availability Is a Data Problem, Not a Logistics Problem →
A step-by-step guide to building a multi-agent framework from scratch, covering planning, tool use, and orchestration without leaning on off-the-shelf agent libraries.
Read more: Agentic AI: Building Autonomous Systems from Scratch →
AI-Powered Customer Support App: Semantic Search with PGVector, Llama2 with an RAG System, and Advanced Translation Models
Read more: AI-Powered Customer Support App →
Step-by-step guide to building visual agents that browse the web autonomously, using multimodal LLMs to reach goals and retrieve information without APIs.
Read more: Building Visual Agents that can Navigate the Web Autonomously →
How Amazon's Chronos adapts language model architectures to zero-shot time series forecasting, with a hands-on benchmark against TiDE on a public dataset.
Read more: Chronos: The Rise of Foundation Models for Time Series Forecasting →
Exploring the intricacies of encoder, multi-head attention, and positional encoding in large language models
Read more: Creating Transformer Encoders and Multi-Head Attention Layers in Python from Scratch →
Document-Oriented Agents: Leveraging ChromaDB, Langchain, and ChatGPT: Enhanced Responses and Cited Sources from Large Document Databases
Read more: Document-Oriented Agents →
Harnessing the Falcon 40B Model, the Most Powerful Open-Source LLM: Mastering open-source language models: diving into Falcon-40B
Read more: Falcon 40B Model, the Most Powerful Open-Source LLM →
How FinalMLP's two-stream MLP with feature gating and bilinear fusion improves CTR prediction, plus a hands-on book recommendation implementation.
Read more: FinalMLP: A Simple yet Powerful Two-Stream MLP Model for Recommendation Systems →
Llama2 for Python Coding Enhancing Llama2’s proficiency in Python through supervised fine-tuning and low-rank adaptation techniques
Read more: Fine-Tune Llama2 for Python Coding on Consumer Hardware →
Open vs Closed Source LLMs, Unpacking Llama 2’s Unique Features, Mastering the Art of Prompt Engineering, and Designing Robust Solutions
Read more: Leveraging Llama 2 Features in Real-world Applications →
Exploring the Transformer’s Decoder Architecture: Masked Multi-Head Attention, Encoder-Decoder Attention, and Practical Implementation
Read more: LLMs and Transformers from Scratch: the Decoder →
Master Semantic Search at Scale: Index Millions of Documents with Lightning-Fast Inference Times using FAISS and Sentence Transformers
Read more: Master Semantic Search at Scale →
Mistral AI vs. Meta: Comparing Top Open-source LLMsA comparison between Mistral 7B vs Llama 2 7B and Mixtral 8x7B vs Llama 2 70B
Read more: Mistral AI vs. Meta: Comparing Top Open-source LLMs →
Build a Bayesian marketing mix model in PyMC to measure channel performance, estimate ROAS, and optimize budget allocation with adstock and saturation.
Read more: MMM: Bayesian Framework for Marketing Mix Modeling and ROAS →
Inside Moirai, Salesforce's universal time series foundation model: its architecture, zero-shot forecasting, and a benchmark against TiDE and Chronos.
Read more: Moirai: Time Series Foundation Models for Universal Forecasting →
How BLIP-2 works and how to combine it with Gemini to build a multimodal fashion search agent that finds products from text and image queries.
Read more: Multimodal Search Engine Agents Powered by BLIP-2 and Gemini →
OpenAI’s Function Calling in Language Learning Models: Transforming Data Pipelines with OpenAI’s Function Calling Feature: Implementing an Email Sending Workflow Using PostgreSQL and FastAPI
Read more: OpenAI’s Function Calling in Language Learning Models →
RAQ, a statistical framework for ranking LLM answer quality, plus a guide to deploying open-source models at scale with NVIDIA NIM microservices.
Read more: Research to Production: Relative Answer Quality (RAQ) and NVIDIA NIM →
A deep dive into tailoring pre-trained LLMs for custom use cases using a RAG approach, featuring LangChain and Hugging Face integration
Read more: Retrieval Augmented Generation →
Mastering Multilingual Audio Transcription and Semantic Search: Unlock the potential of cross-language information accessibility
Read more: Mastering Multilingual Audio Transcription and Semantic Search →
A walkthrough of Meta's Seamless translation family — SeamlessM4T v2, Expressive, and Streaming — with architectures and a Python guide using HuggingFace.
Read more: Seamless: In-Depth Walkthrough of Meta’s New Open-Source Suite of Translation Models →Stable Diffusion: Mastering the Art of Interior Design: A deep dive into Stable Diffusion and its inpainting variant for interior design
Read more: Stable Diffusion: Mastering the Art of Interior Design →
Speech Model that Supports 1162 Languages: Explore the cutting-edge multilingual features of Meta’s latest automatic speech recognition model
Read more: Testing the Massively Multilingual Speech Model →
TFT: an Interpretable Transformer. Graphic representation of TFT, an interpretable transformer, showcasing its architecture and features.
Read more: TFT: an Interpretable Transformer →
How Llama's architecture evolved from Llama 1 to 3.1 — GQA, tokenizer, and context changes — plus a hands-on Llama 2 vs Llama 3 comparison using RAQ.
Read more: The Evolution of Llama: From Llama 1 to Llama 3.1 →A deep exploration of TiDE, its implementation using Darts and a real life use case comparison with DeepAR and TFT
Read more: TiDE: the ‘embarrassingly’ simple MLP that beats Transformers →
TimeGPT vs TiDE: Is Zero-Shot Inference the Future of Forecasting or Just Hype?
Read more: TimeGPT vs TiDE →
TimesFM, Google's time series foundation model trained on 307B points: architecture, zero-shot inference, and benchmarks vs TiDE, Chronos, and MOIRAI.
Read more: TimesFM: The Boom of Foundation Models in Time Series Forecasting →
Whisper JAX vs PyTorch: Uncovering the Truth about ASR Performance on GPUs
Read more: Uncovering the Truth about ASR Performance on GPUs →
Unlock the Power of Audio Data: Advanced Transcription and Diarization with Whisper, WhisperX, and PyAnnotate
Read more: Unlock the Power of Audio Data →
GPT-4V and Text-to-Speech Technology Enhancing Visual Impairment Navigation: Integrating GPT-4V and TTS for Advanced Sensory Assistance
Read more: Visually Impaired with GPT-4V and Text-to-Speech Technology →
Vosk for Efficient Enterprise-Grade Speech Recognition: A comprehensive walkthrough of implementing and assessing Vosk-based speech recognition systems
Read more: Vosk for Efficient Enterprise-Grade Speech Recognition →Tell us about the decision you're trying to make. We'll tell you whether Horizon, Pliney, or a Custom AI Solution fits, or whether it's none of the above.