Skip to content
datarekha
Blog Production AI · GATE DA · how concepts actually work · Page 11 of 12

From theory to the systems that demand real understanding.

Long-form pieces on how production-AI teams orchestrate agents, serve models, and build the stack, plus GATE DA essays that turn probability, linear algebra, DBMS, ML, and AI into durable concepts.

281 posts · 20 categories
Cloud May 12, 2026

Cloud Run is the most underrated platform for AI agents

Everyone deploys agents on GKE or EC2. The teams quietly shipping the fastest, cheapest production agents have moved to Cloud Run + Gemini + Firestore — and the reasons are not what you'd guess from the marketing pages.

11 min read Read
RAG May 12, 2026

Hybrid search, when one retriever isn't enough

Pure vector search misses exact identifiers. Pure BM25 misses paraphrases. Every serious production RAG stack — Notion, Perplexity, Vespa, Glean — runs both and fuses the results. Here's the math, the latency budget, and the production knobs that matter.

12 min read Read
Infrastructure May 12, 2026

vLLM vs TGI vs SGLang: choosing your inference server in 2026

Three open-source serving stacks, three different bets. vLLM optimises raw throughput. TGI optimises ease and HuggingFace integration. SGLang optimises structured generation and prefix-cache reuse. Here's how to pick.

13 min read Read
Agents May 11, 2026

Long-horizon agent failure modes: context rot, drift, looping

An agent that handles ten turns gracefully will likely collapse at fifty. The reasons aren't bugs — they're emergent properties of how transformers handle growing context. Here's a field guide to how agents disintegrate, with the public failure stories that mapped the territory.

13 min read Read
Frameworks May 11, 2026

Microsoft Agent Framework, six months in

Microsoft consolidated Semantic Kernel and AutoGen into a single Agent Framework in late 2025. Half a year later, the picture is clear: MAF is the Azure-native option for shops that already speak C# and .NET, and its workflow/agent split is the most pragmatic in the industry.

11 min read Read
Agents May 10, 2026

Coding agents in 2026: Cursor, Devin, Sweep, Aider, Claude Code compared

Two years after Devin launched and froze the term 'AI software engineer' in the popular imagination, here's where everyone actually landed. The dominant tools didn't converge — they specialized, and the ones that won did so by picking a workflow lane and dominating it.

14 min read Read
RAG May 10, 2026

Contextual Retrieval, two years on

Anthropic's September 2024 trick — prepend a sentence of context to each chunk before embedding — quietly became the new RAG default. Here's why a 49% retrieval-error reduction held up, what the numbers actually say, and when you still shouldn't bother.

11 min read Read
Agents May 9, 2026

Deep research agents explained: Perplexity, GPT Deep Research, Gemini Deep Research

Deep Research was OpenAI's first agent product to actually ship, and it landed in a market where Google had already been live for months and Perplexity had built the search-citation pipeline the others were chasing. The three converged on similar capabilities but radically different architectures.

13 min read Read
Infrastructure May 9, 2026

How Anthropic serves a hundred million tokens a second

The frontier labs and their serving partners — Anthropic, OpenAI, Together AI, Fireworks — hide an arsenal of inference optimisations behind the simple-looking chat endpoint. Here's the hierarchy, what each layer bought, and the public numbers behind it.

13 min read Read
Agents May 8, 2026

Agentic data analysis: PandasAI, Hex Magic, Julius — and why it's harder than it looks

'Just talk to your data' has been the cleanest agentic AI demo for two years. The shipped reality is much narrower than the demos suggest, because the gap between answering a question and answering it cheaply, safely, and correctly is enormous. Here's where the production line lives.

13 min read Read
Frameworks May 8, 2026

LangGraph is what you reach for when the agent has to remember

By mid-2026 LangGraph is the default orchestration layer for stateful, long-running, human-in-the-loop agents. Not because of the API — because of the checkpoints, the interrupts, and the fact that the state outlives the request.

11 min read Read
Agents May 7, 2026

The agent observability stack: LangSmith, Langfuse, Helicone, Arize

You can't ship agents without observability. The 2026 stack has settled into three layers — traces, evals, drift — and five vendors who each won by being unmistakably the best at one of them. Here's who picks which, and why.

13 min read Read
Agents May 6, 2026

Agents for legal, finance, healthcare — the high-stakes pattern

Three industries where hallucination is malpractice. The agent companies winning in legal, finance, and healthcare have converged on the same shape — scoped tasks, retrieval over generation, mandatory human checkpoints, audit logging by default. Here's what that pattern looks like in production at Harvey, Hebbia, Abridge, and Suki.

14 min read Read
ML May 5, 2026

Why XGBoost is still winning in 2026

Twelve years after the original paper, and despite a parade of transformer-based challengers, gradient-boosted trees remain the default for tabular data at most enterprises. Here's why the boring answer keeps winning — and what TabPFN, CatBoost, and LightGBM actually changed.

13 min read Read
Infrastructure May 4, 2026

MLOps platform consolidation: Databricks, Snowflake AI, SageMaker

The MLOps landscape of 2023 — a dozen point tools, three feature stores, four monitoring vendors, two training frameworks — has collapsed. By 2026 the workload lives on Databricks, Snowflake Cortex, SageMaker Unified Studio, or Vertex. Here's what each platform actually does, what got swept under, and who wins which workload.

13 min read Read
Infrastructure May 3, 2026

Feature stores in 2026: Tecton, Feast, Hopsworks — death and rebirth

The feature store hype cycle went peak (2021), trough (2023, 'just use dbt'), and into a quieter rebirth as the data-for-AI layer for both classical ML and LLM agents. Tecton sold to Databricks. Feast survives as the open-source baseline. Hopsworks redefined itself as an AI lakehouse. Here's what the modern feature store actually is in 2026.

13 min read Read
Infrastructure May 2, 2026

The vector database shakeout: Pinecone, Weaviate, Qdrant, Chroma, pgvector

The pure-play vector DB category isn't dead, but it's smaller than the 2023 hype suggested. Postgres ate the low end, Pinecone owns serverless ops, Qdrant is the open-source cost winner, and Weaviate's hybrid search is its real moat. Here's the 2026 picture by the numbers.

12 min read Read
RAG May 1, 2026

Embeddings in 2026: Voyage, Cohere v4, OpenAI text-embedding-3, mxbai, BGE

The embedding model market split clean. Proprietary models (Voyage 3, Cohere v4, OpenAI v3) own the quality top; open-source (mxbai, BGE-M3, Jina v3, E5) own cost and self-host. Matryoshka rewrote the cost economics. Here's how to actually choose.

12 min read Read
RAG Apr 30, 2026

Reranking at scale: cross-encoders, ColBERT, mxbai

Retrieve-then-rerank is the single biggest accuracy improvement in production RAG. Three approaches dominate — cross-encoders, late interaction, API rerankers. Here's what each costs, what each wins, and the latency budget that decides between them.

12 min read Read
Infrastructure Apr 29, 2026

Distributed training: FSDP vs DeepSpeed vs Megatron in production

For serious pretraining or fine-tuning, you pick from three: FSDP2 (PyTorch native), DeepSpeed (ZeRO stages), or Megatron-LM (NVIDIA, 3D parallel). The frontier labs have made their bets — Llama 3 went Megatron, FSDP2 is the open-source default under 70B, and DeepSpeed survives where ZeRO offload is necessary. Here's how to choose.

13 min read Read
Infrastructure Apr 28, 2026

Mixture of experts in production: Mixtral, DeepSeek, Llama 4

MoE went from research curiosity in 2021 to production default in 2025. The new frontier-model template is sparse, with 5-15% active parameters per token. Here's how the serving infrastructure caught up, what vLLM and SGLang actually do with experts, and the cost math that makes MoE the future.

13 min read Read
Infrastructure Apr 27, 2026

Quantization in production: GPTQ, AWQ, GGUF, FP8 — what to ship

16-bit serving is dead for most production workloads. Here's how to pick between weight-only post-training quantization, activation-aware quantization, and the FP8 native path — without measuring it wrong.

12 min read Read
MLOps Apr 26, 2026

ML platform build vs buy: a decision framework for 2026

Most teams default to one of three answers — Databricks, SageMaker, or 'we'll roll our own.' All three are wrong for the wrong team. Here's how to pick by team size, model variety, and latency budget — with the war stories that explain why.

13 min read Read
MLOps Apr 25, 2026

Model monitoring in 2026: from accuracy to behavior drift

Classical ML monitoring tracks accuracy decay. LLM monitoring tracks something stranger — the model itself silently changing underneath you. Here's what production observability looks like when the failure modes don't fit the dashboards you built five years ago.

12 min read Read