Start returning runners with baseline plyometrics and short-distance running
For someone who hasn't run in years, begin with very basic baseline plyometrics and keep all running to very short distances.
37,617 results · Most relevant
For someone who hasn't run in years, begin with very basic baseline plyometrics and keep all running to very short distances.
Fix 1: stay forward enough for longer — standing up immediately after initiating a sprint means you can't get enough horizontal force into the ground. Fix 2: take more steps in the initial 5-10 meters instead of huge semi-lunge strides, even though long strides feel like pushing hard. Fix 3: keep on the train track — avoid lateral sway and driving out to the side each step, especially for bigger athletes.
What is important is to match your bottleneck rather than simply going with a lever that has worked before. For this pipeline the speaker chooses async IO for fetching images and Ray Actors as the multiprocessing library for processors. Ray Actors are essentially the same as general processes but offer an API to communicate across workers, so you don't have to reinvent the wheel.
Disaggregate the process so workers are spread across multiple nodes, doing their own image processing and storing, retrieved lazily just-in-time by DP ranks. Two flags to tune for disaggregation: zero-copy retrieval and a spread scheduling mechanism.
Spread data-processing workers across multiple nodes so they do their own image processing and storing, with results retrieved lazily just-in-time by data parallel ranks.
Use a background thread that fills up a queue ahead of time and always stays ahead of the consumer, so the trainer is never bottlenecked on requesting data. The producer should produce examples at a multiplier of the consumer's rate; making sure your producer is always ahead of your consumer will pay great dividends.
Use an object store (a Ray convenience method) where each sample carries an object reference (a pointer to the data) instead of the data itself. With an object store, the tensor never round trips to the head node, and Megatron DP ranks can directly call the array store to fetch data as needed. The Ray store also eliminated an extraneous put object call that the prefetcher solution had.
You need a syncing layer on top of MCP live lookups for anything you want to semantically question or go deeper into — sync a copy of the data locally into a vector database, which AI can build relatively easily. If you need any form of semantic lookup or analysis beyond a simple lookup — anything requiring looking across a vast data set — you have to sync a copy. Static, single-entity lookups (e.g., asking Stripe a question about a specific customer or taking a single action) can remain live MCP calls.
Pedro Lopez recommends OAuth for authenticating the MCP server itself: a connect button spins up the OAuth flow where the user signs in and accepts privileges.
After trial and error, Pedro Lopez's team keeps tool counts small, exposing only describe (schema for querying) and execute (read/write/delete) for connector execution.
Pedro Lopez recommends exposing CLI inputs through flags, environment variables, and config files instead of interactive prompts, so setup can happen out of band.
Anders Swanson recommends using a database for the memory layer rather than building your own database on top of files: building a memory layer on files leads to building your own database, so you could just use a database instead. With one database query connection in a multi-model database, you can do hybrid search and return the data back.
endorsed
Anders Swanson recommends choosing a multi-model database for the memory layer so rows, columns, vector embeddings, full text search, relational entities, graphs, episodes, and facts live in one consolidated place.
endorsed
Viren Baraiya recommends that for payments you use a very well-defined deterministic workflow rather than non-deterministic agent behavior. Similarly, for an agent harness responsible for monitoring production deployment and managing Kubernetes clusters, define a very well-defined workflow for the sequence of steps to execute when restarting a cluster, so that every cluster restart is exactly the same. The harness should deliver determinism when it matters.
Viren Baraiya recommends that restarting a production cluster should require a human gate: send a Slack message to somebody saying 'I'm going to restart this cluster, do you think it's okay to do that or not?' He does not want this to be left to a hallucination by an LLM — execution guarantees must be deterministic, not left to an LLM deciding it doesn't need to do it.
Gabriel Martinez recommends that engineers read the code and test the behavior themselves rather than accepting output, own the trade-offs of the code they merge, and notice the missing requirements and decide whether a thing deserves to be part of the system. He also recommends not merging code you don't fully understand or haven't tested yourself.
Creator advice attributed to Gabriel Martinez.
Gabriel Martinez recommends spending time thinking about guardrails for agents, coding conventions, architectural styles, and constraints around how the system should be shaped. He frames this as fighting slop with conventions, standards, documented flows, small reviewable surfaces, and clear ownership boundaries and definitions of done.
Creator advice attributed to Gabriel Martinez.
Postman built an API context graph by: (1) identifying and cataloging every single microservice on production, (2) identifying every endpoint exposed by each REST API service, (3) capturing how each endpoint is implemented right down to the exact line of code, (4) mapping how each service talks to each other and how each endpoint connects to every other endpoint, (5) tracking how data flows all the way down across the stack including databases and caches, (6) identifying how applications like front-ends, CLIs, and APIs interact with the back-end architecture, and (7) indexing all of this data using an LLM to produce a centralized and effective context graph layer.
endorsed
When a PR is created, the agent reviews it and posts issues and inline comments; it also analyzes CI failures and posts a summary of root causes, being particularly good at catching flaky tests; the agent can automatically fix code review issues and CI failures, and can be set up to automatically loop until all issues in a PR are fixed, producing a green PR ready to merge.
Will Lyon recommends running agent messages (user and assistant) through an entity extraction process that identifies entities and how they are connected, converting unstructured data into a knowledge graph.
Choose sources → to get a library of your own and a daily brief.