Committed to sharing tech and documenting ideas.
The article traces OpenAI’s seven‑year journey from the OpenAI Five Dota 2 victory to ChatGPT, showing how a clear goal, massive self‑play, scaling of compute and data, and systematic transfer of learned capabilities enabled the transition from game‑playing AI to a widely used conversational product.
The article traces four generations of enterprise data intelligence—from BI dashboards to Text‑to‑SQL and finally Data Agents—explaining why successful deployment depends on five engineering disciplines (Context, Knowledge, Skill, Memory, Evaluation Loop) rather than the underlying LLM model.
This article explains why and how knowledge distillation lets a 1.5 B parameter model surpass much larger LLMs on math benchmarks, detailing the underlying soft‑label transfer, temperature tuning, various distillation families, engineering pipelines, and the practical trade‑offs that bound its success.
The article dissects Cursor's dual‑index architecture—semantic vector search built on AST chunking and a custom‑trained embedding model, plus a local trigram regex index—explaining how Merkle trees, team index reuse, and file‑based context delivery enable sub‑second code retrieval in massive monorepos.
The article analyzes why agents often forget or hallucinate past information, defines three orthogonal memory dimensions, critiques common naive solutions, and presents a four‑module architecture—including extraction, representation, retrieval, and maintenance—plus cross‑cutting concerns, a reference design, failure patterns, and a workload‑driven selection matrix.
The article compares three workflow execution models—traditional BPM token machines, Dify's event‑driven graph scheduler, and LLM‑based Agent loops—detailing their core data structures, decision‑making mechanisms, concurrency models, error handling, and how each shifts decision authority from developers to AI.
The article presents a six‑step AI‑driven system that transforms casual queries into a disciplined learning process by defining a learning path, testing knowledge, compressing material into quick reference sheets, curating high‑impact resources, and applying the Feynman loop for deep mastery.
Loop Engineering extends the ReAct inner loop by adding an outer control system that automates task discovery, scheduling, verification, and termination, turning AI agents from manual schedulers into self‑directed machines, while highlighting necessary components, risks, and best‑practice guidelines for robust, scalable AI‑driven development.
The article dissects Cursor's Agentic programming platform, revealing that performance gains stem from four engineering layers—architecture, inference, transport, and Agent loop—rather than merely larger models, and offers concrete optimization techniques and lessons for building fast, reliable cloud agents.
The article examines how a single engineer can incur $211 in daily AI coding fees, explores Uber's budget blowout, explains the quadratic token cost of large context windows, and offers a step‑by‑step framework for small‑to‑mid‑size companies to measure, control, and justify AI tool spending.
