Lessons

The two transistor families that built modern electronics. BJTs as current-controlled amplifiers, MOSFETs as voltage-controlled switches with a capacitor for a gate, the three operating regions of each, and the four reasons MOSFETs ended up running every digital chip.

From a single PN junction to the four-diode bridge in every wall adapter. The IV curve, rectification, Zener voltage references, fast Schottky diodes for switching supplies, LEDs and photodiodes turning current to light and back, and the five failure modes that bite real designs.

What makes silicon work where diamond won't, why doping turns an insulator into a tunable conductor, and how slapping p-type silicon next to n-type creates the depletion region that becomes a diode. The physics every chip is built on.

Make loops trustworthy. Adversarial verification panels, sub-agent orchestration, loop-until-dry for unbounded discovery, loop-until-budget for paid depth, multi-modal sweeps with diverse prompts, eval-driven cap selection, and the five anti-patterns — silent caps, infinite plans, drift, premature termination, thrashing — that ship to prod more than they should.

Once an agent loop runs past a few iterations the failure surface shifts: state bloats, tool calls fail in three distinct ways, retries multiply if you stack them wrong, and unbounded budgets rack up four-figure bills overnight. The state, errors, and budgets you need to make the loop survive contact with reality.

How LLM agents actually run: the iterative prompt-action-observation loop, the ReAct shape, the smallest tool-calling loop in twelve lines, why you always stack three termination layers, what the model sees on iteration N, and when a single prompt is the better answer.

Master the OTel GenAI semantic conventions — gen_ai.* attributes, span structure for prompts/completions/tools, sampling strategies, and cost attribution — and understand why standardizing across LangSmith, Phoenix, Datadog, and Grafana matters for production AI systems.

Learn how to systematically evaluate Retrieval-Augmented Generation systems in production using RAGAS, TruLens, and Phoenix — covering golden sets, retrieval drift, embedding drift, and cost-aware eval scheduling.

A rigorous look at how to measure agent performance — trajectory-level vs final-answer evals, canonical multi-step benchmarks (SWE-bench, WebArena, OSWorld, GAIA), LLM-as-judge pitfalls, and why your eval is your spec.

A deep look at how reinforcement learning on chains-of-thought powers o1, DeepSeek-R1, Claude reasoning, and Gemini Thinking — covering GRPO, MCTS-style search, test-time compute scaling, and distillation into smaller models.

How Group Relative Policy Optimization turns unit tests and theorem provers into training signals — the technique behind DeepSeek-R1's leap in math and code reasoning without a reward model.

A deep dive into DPO (Rafailov et al. 2023) and its successors — how they reformulate RLHF as a classification problem, the math behind the implicit reward, and where each variant wins or loses against PPO-based pipelines.

Trace the full arc of Reinforcement Learning from Human Feedback — preference data, Bradley-Terry reward models, PPO with KL penalty, reward hacking, and why most labs have moved to offline alternatives like DPO and RAFT.

An honest map of RL in 2026 — the domains where it actually reaches production (LLM post-training, robotics policies, ad bidding, RLHF, reasoning models) and the places where it still cannot reliably cross the lab-to-deployment gap.

A deep dive into how large language models actually run in production — why prefill is fast and decode is slow, how the KV cache works, sampling strategies like temperature and top-p, speculative decoding, and continuous batching with vLLM.

A deep dive into how raw pretrained language models become helpful assistants — from supervised fine-tuning on curated demonstrations, through reward modeling and PPO-based RLHF, to modern direct alignment methods like DPO and the recipes used in Llama 3, Llama 4, and DeepSeek.

How two landmark papers — Kaplan et al. 2020 and DeepMind's Chinchilla 2022 — rewrote our understanding of compute-optimal training, why the industry now deliberately overtrains models, and how inference costs flip the math entirely.

A deep tour of the architectural choices powering today's large language models — decoder-only Transformers, encoder-decoder designs, grouped-query attention, RoPE positional embeddings, and mixture-of-experts routing — with concrete numbers and trade-offs.

A deep dive into how large language models learn from raw text: the next-token prediction objective, cross-entropy loss, the messy reality of web data curation (Common Crawl, dedup, quality filters), and the lineage from The Pile to FineWeb.

A rigorous tour of how modern LLMs split text into tokens — covering byte-pair encoding, GPT-2/Llama's byte-level variant, SentencePiece, vocabulary design trade-offs, and why your tokenizer silently determines multilingual fairness, code quality, and arithmetic ability.

How to manage, compress, and strategically fill the context window in long-horizon agents — covering summarization checkpoints, scratchpad memory, retrieval injection, prompt caching, and compaction triggers.

A deep dive into designing tool schemas that LLMs actually call correctly — covering parameter naming, description quality, structured output via response schemas, output parsing, and error message design that drives self-correction.

A practical deep-dive into fan-out, pipeline staging, judge panels, adversarial verification, and loop-until-dry — plus an honest accounting of when spawning subagents actually earns its token cost and when it just burns money.

Move beyond toy ReAct loops. Learn how to build a production-grade agent harness with a robust control loop, tool registry, schema validation, retry logic, token budgets, abort signals, and a persistent journal that survives crashes.