What did @kirkdborne post on X, week of Sep 15 – Sep 21, 2026?

AI, LLMs & Agentic Systems

Agentic AI and multi-agent system design dominated the week, with @kirkdborne repeatedly highlighting books on production agent architectures — from single-agent deployments to multi-agent orchestration using frameworks like LangGraph, CrewAI, and MCP [1], [2], [3], [4], [5]. On LLMs and RAG, he pushed resources covering fine-tuning techniques (LoRA, QLoRA, DPO), retrieval-augmented generation pipelines, and inference optimization methods like KV Cache and Paged Attention [6], [7], [8], [9], [10], [11], [12], [13]. For GenAI applications and AI career development, he recommended full-stack guides for building with Claude, Gemini, and OpenAI, alongside strategic AI leadership playbooks [14], [15], [16], [17], [18], [19], [20], [21]. He also surfaced AI systems, security, and MLOps resources — including the Harvard ML Systems curriculum, AI networking automation, and adversarial AI threat modeling guides [22], [23], [24], [25], [26], [27], [28].

Agentic AI & Multi-Agent Systems: He spotlighted books on building production-ready agents, including '30 Agents Every AI Engineer Must Build' and 'The Most Complete AI Agentic Engineering System,' covering memory, planning, reasoning, guardrails, and multi-agent orchestration [29], [1], [2], [4], [5]. He promoted a Packt workshop on deploying AI agents with a 40% discount code [30], [31].

LLMs, RAG & Fine-Tuning: He recommended a Super Study Guide for Transformers and LLMs with ~600 illustrations, a Stanford LLM course, and deep-dives into inference optimization (KV Cache, Flash Attention, Speculative Decoding) [7], [10], [32], [33]. He shared resources on RAG pipelines, GraphRAG, and fine-tuning techniques like LoRA, QLoRA, and VeRA [6], [8], [9], [13].

GenAI Applications, Tools & Career: He pushed practical guides for building with Claude, Gemini, and OpenAI APIs, including 'Build AI-Enhanced Web Apps' and the 'Claude Visual Bible' [14], [16], [17], [18], [34], [35], [36], [37], [38], [39]. He also highlighted AI leadership and career books like the 'Chief AI Officer Handbook' and 'Generative AI Career Masterplan' [15], [40], [41], [42], [19], [43], [44], [20], [45].

AI Systems, Security & MLOps: He shared the Harvard ML Systems curriculum (28K GitHub stars) as a free alternative to paid AI engineering courses, alongside books on AI networking, MLOps, and system design [22], [23], [46], [47], [26]. He flagged AI security threats, recommending books on adversarial AI attacks, LLM security, and agentic cybersecurity [48], [49], [25], [27], [28].

Mathematics

Math education remained a core theme, with @kirkdborne sharing free PDFs on linear algebra, discrete math, and game theory alongside books on mathematical puzzles and number curiosities [50], [51], [52], [53], [54], [55]. He repeatedly argued that math fear is planted, not innate, recounting a true story of teaching calculus to students without telling them it was calculus until after their first quiz — 'some had tears in their eyes' [56]. He tied this to his own career pivot from astrophysics to ML after reading Tom Mitchell's definition of ML as 'mathematical algorithms that learn from experience' [57], [58]. He recommended foundational texts for AI/ML math, including 'The Mathematics of Large Language Models' and 'Mathematics for AI and Machine Learning,' and gave a detailed personal review of Tivadar Danka's 'Mathematics of Machine Learning' [59], [60], [61], [62], [63], [64], [58], [65].

Data Science & Machine Learning

He shared classic and modern ML textbooks, including a free 421-page PDF of Tom Mitchell's 'Machine Learning' and the comprehensive 'Deep Learning' textbook by Goodfellow et al. [66], [67], [57], [68], [69], [70], [71], [72], [73]. For data engineering and analytics, he recommended books on Tableau, Streamlit, database design, and managing data as a product [74], [75], [76], [77], [78], [79], [80], [81], [82], [83], [84]. He also highlighted Bayesian modeling, causal inference resources from Judea Pearl, and a Kaggle competition guidebook that he reviewed as a '700-page masterpiece' [85], [86], [87], [88], [89], [90], [91], [92], [93], [94], [81].