---
title: "Harpreet Sahota (@datascienceharp) | Hashnode"
description: "Hacker-in-Residence at The Employable Data Scientist. Read the latest articles by Harpreet Sahota on Hashnode."
url: "https://hashnode.com/@datascienceharp"
---

@datascienceharp

# Harpreet Sahota

@datascienceharp · Joined October 2026

Hacker-in-Residence at The Employable Data Scientist

[Website](https://employabledatascientist.com/)

### [HapTile: 1,699 UR5e Episodes With Tactile Video and Haptic Levels, in FiftyOne](https://hashnode.employabledatascientist.com/haptile-fiftyone-analysis?utm_source=hashnode&utm_medium=feed)

HapTile pairs UR5e teleoperation with two fingertip tactile streams, two RGB-D cameras and haptic levels. Scrub 1,699 episodes in FiftyOne.

6 min read · [hashnode.employabledatascientist.com](https://hashnode.employabledatascientist.com/haptile-fiftyone-analysis?utm_source=hashnode&utm_medium=feed)

[Nov 6](https://hashnode.com/posts/haptile-fiftyone-analysis/6af0923ef375e93ad330a682)

### [Building a Data Quality Feedback Loop for Visual AI Systems](https://hashnode.employabledatascientist.com/data-quality-feedback-loop-visual-ai?utm_source=hashnode&utm_medium=feed)

Stop wasting engineering sprints tuning hyperparameters on flawed datasets. Here is how to build a 4-step data-centric evaluation loop for computer vision.

7 min read · [hashnode.employabledatascientist.com](https://hashnode.employabledatascientist.com/data-quality-feedback-loop-visual-ai?utm_source=hashnode&utm_medium=feed)

[Nov 6](https://hashnode.com/posts/data-quality-feedback-loop-visual-ai/6aedf151bf50f661bd3f03be)

### [A 3-Step Data Quality Pre-Flight Check Before Training Vision Models](https://hashnode.employabledatascientist.com/three-step-data-quality-preflight-check?utm_source=hashnode&utm_medium=feed)

New architectures drop every 33 hours, but training success depends on data fundamentals. Run this 3-step audit on resolution, labels, and class balance.

6 min read · [hashnode.employabledatascientist.com](https://hashnode.employabledatascientist.com/three-step-data-quality-preflight-check?utm_source=hashnode&utm_medium=feed)

[Nov 6](https://hashnode.com/posts/three-step-data-quality-preflight-check/6aef3687291dbbcf87f2441b)

### [Critical Image Quality Metrics That Break Computer Vision in Production](https://hashnode.employabledatascientist.com/critical-image-quality-metrics-computer-vision?utm_source=hashnode&utm_medium=feed)

Vision models fail when real-world camera conditions diverge from training data. How to measure and audit blur, exposure, contrast, and compression artifacts.

6 min read · [hashnode.employabledatascientist.com](https://hashnode.employabledatascientist.com/critical-image-quality-metrics-computer-vision?utm_source=hashnode&utm_medium=feed)

[Nov 6](https://hashnode.com/posts/critical-image-quality-metrics-computer-vision/6af087ad9b5a2f02b53ea6f0)

### [Why Squeaky-Clean Datasets Break Production Vision Models](https://hashnode.employabledatascientist.com/why-squeaky-clean-datasets-fail-in-production?utm_source=hashnode&utm_medium=feed)

Over-cleaning training data strips structured real-world noise, leaving vision models brittle. Build a data preservation list before running cleaning sprints.

5 min read · [hashnode.employabledatascientist.com](https://hashnode.employabledatascientist.com/why-squeaky-clean-datasets-fail-in-production?utm_source=hashnode&utm_medium=feed)

[Nov 6](https://hashnode.com/posts/why-squeaky-clean-datasets-fail-in-production/6af09254f375e93ad330cf8d)
