Show HN: Fitness MCP
https://getfast.ai/claude-connector
There's no external MCP for your fitness (Garmin / Strava) data, so we built one.
tmulc18
Nano Banana on Gemini is not the same as on AI Studio
Re:Ask HN: Who is hiring? (December 2025)
on Monday December 01, 2025 @01:22PM (#46110931)
GetFast | Founding Engineer | Bay Area / Remote
We're on a mission to simplify running through wearables data. We're a small team of runners who love to build.
We're looking for a founding engineer to shape the future of our product and company. You’ll work across the stack–from building reliable, scalable systems to rapidly prototyping and launching new user-facing features. You’ll have significant ownership and influence on architecture, product direction, and engineering culture.
Ideally you love a few of the following: - Foundation ML models in the sensory space - Chat agents - Novel data visualizations - Recommendation systems - Operations research - Unique UX
Re:Ask HN: What Are You Working On? (Nov 2025)
on Sunday November 09, 2025 @05:58PM (#45870096)
https://getfast.ai/strava/login
Using transformers to understand fitness data
Re:Race SPT: Sensor Pretrained Transformer
on Thursday October 30, 2025 @07:15PM (#45766534)
We trained a model on sensor data to learn useful patterns for predicting running race performances.
Race SPT: Sensor Pretrained Transformer
Re:Everything Is Correlated
on Friday August 22, 2025 @02:49AM (#44981658)
gwern is goated
Show HN: Chat with Your Wearables Data
I stumbled into the effective oxygen percentage by looking at model errors
Show HN: I built a tool to show how people trained for your goal marathon time
The idea for this came from wanting something like Naviance (let’s you see the GPA and test scores of admitted students for a given college) for marathon running.In general, we’re working on making your running data more useful, and our newest tool, called, "Marathon Cohorts," gives you stats about runners who finished a marathon in a given time. You pick a marathon time / cohort and then look...
Re:Show HN: MidWord – A Word-Guessing Game
on Wednesday June 11, 2025 @01:25PM (#44249803)
Would be more fun (even if you call it "easy mode") if you had the alphabet at the top of the page! Took me 17 tries to get "prepare" but I had a lot of mistakes because I forgot what came after what in the alphabet.
Re:Tensorflow 2.0: models migration and new design
on Tuesday November 13, 2018 @05:57PM (#18445700)
2.5 years of learning tensorflow now gone to waste :(
Vanilla web app for labeling audio files
Show HN: Microsoft releases AirSim
Distributed TensorFlow (DTF) Guide
Re:Speech Recognition Is Not Solved
on Tuesday October 24, 2017 @07:46PM (#15546181)
In the caption for figure 2 there is this quote, "Notice the humans are worse at transcribing the non-American accents." But when I look at the bar graph, I see the word error rates of the model is much higher than the word error rates of humans for most of the non-American accents. Am I missing something?
Deep Voice 3: 2000-Speaker Neural Text-To-Speech
Distributed Deep Learning Reading List
USPO grants Google a patent on dropout
Re:How Booking.com manipulates users
on Wednesday September 20, 2017 @06:43PM (#15299028)
That's why you gotta use Expedia or Google
Re:Advantages of Using R Notebooks for Data Analysis Instead of Jupyte...
on Friday June 09, 2017 @06:21PM (#14525042)
Really cool article. It still seems like jupyter is the better longterm option because it offers so many different kernels.
Re:Reconstructing Artistic Style
on Friday May 05, 2017 @02:11PM (#14275545)
It's much easier now since people have had time experimenting with hyperparameters. But I've never implemented deep dream so idk how difficult that is