Picture by Creator
# Introduction
Though in fashionable information science you’ll primarily discover Jupyter notebooks, Pandas, and graphical dashboards, they don’t all the time provide the degree of management you may want. Alternatively, command-line instruments might not be as intuitive as you would like, however they’re highly effective, light-weight, and far quicker at executing the particular jobs they’re designed for.
For this text, I’ve tried to create a stability between utility, maturity, and energy. You’ll discover some classics which might be practically unavoidable, together with extra fashionable additions that fill gaps or optimize efficiency. You’ll be able to even name this a 2025 model of essential CLI instruments listing. For individuals who aren’t aware of CLI instruments however need to study, I’ve included a bonus part with assets within the conclusion, so scroll all the best way down earlier than you begin together with these instruments in your workflow.
# 1. curl
curl is my go-to for making HTTP requests like GET, POST, or PUT; downloading recordsdata; and sending/receiving information over protocols resembling HTTP or FTP. It’s best for retrieving information from APIs or downloading datasets, and you’ll simply combine it with data-ingestion pipelines to tug JSON, CSV, or different payloads. One of the best factor about curl is that it’s pre-installed on most Unix techniques, so you can begin utilizing it straight away. Nevertheless, its syntax (particularly round headers, physique payloads, and authentication) may be verbose and error-prone. When you find yourself interacting with extra advanced APIs, it’s possible you’ll choose an easier-to-use wrapper or Python library, however realizing curl continues to be a vital plus for fast testing and debugging.
# 2. jq
jq is a light-weight JSON processor that allows you to question, filter, rework, and pretty-print JSON information. With JSON being a dominant format for APIs, logs, and information interchange, jq is indispensable for extracting and reshaping JSON in pipelines. It acts like “Pandas for JSON within the shell.” The largest benefit is that it supplies a concise language for coping with advanced JSON, however studying its syntax can take time, and very giant JSON recordsdata could require further care with reminiscence administration.
# 3. csvkit
csvkit is a set of CSV-centric command-line utilities for reworking, filtering, aggregating, becoming a member of, and exploring CSV recordsdata. You’ll be able to choose and reorder columns, subset rows, mix a number of recordsdata, convert from one format to a different, and even run SQL-like queries in opposition to CSV information. csvkit understands CSV quoting semantics and headers, making it safer than generic text-processing utilities for this format. Being Python-based means efficiency can lag on very giant datasets, and a few advanced queries could also be simpler in Pandas or SQL. If you happen to choose pace and environment friendly reminiscence utilization, contemplate the csvtk toolkit.
# 4. qwk / sed
Hyperlink (sed): https://www.gnu.org/software program/sed/handbook/sed.html
Traditional Unix instruments like awk and sed stay irreplaceable for textual content manipulation. awk is highly effective for sample scanning, field-based transformations, and fast aggregations, whereas sed excels at textual content substitutions, deletions, and transformations. These instruments are quick and light-weight, making them good for fast pipeline work. Nevertheless, their syntax may be non-intuitive. As logic grows, readability suffers, and it’s possible you’ll migrate to a scripting language. Additionally, for nested or hierarchical information (e.g., nested JSON), these instruments have restricted expressiveness.
# 5. parallel
GNU parallel quickens workflows by operating a number of processes in parallel. Many information duties are “mappable” throughout chunks of information. Let’s say you must execute the identical transformation on tons of of recordsdata—parallel can unfold work throughout CPU cores, pace up processing, and handle job management. It’s essential to, nevertheless, be conscious of I/O bottlenecks and system load, and quoting/escaping may be difficult in advanced pipelines. For cluster-scale or distributed workloads, contemplate resource-aware schedulers (e.g., Spark, Dask, Kubernetes).
# 6. ripgrep (rg)
ripgrep (rg) is a quick recursive search software designed for pace and effectivity. It respects .gitignore by default and ignores hidden or binary recordsdata, making it considerably quicker than conventional grep. It’s good for fast searches throughout codebases, log directories, or config recordsdata. As a result of it defaults to ignoring sure paths, it’s possible you’ll want to regulate flags to look every little thing, and it isn’t all the time obtainable by default on each platform.
# 7. datamash
datamash supplies numeric, textual, and statistical operations (sum, imply, median, group-by, and many others.) instantly within the shell through stdin or recordsdata. It’s light-weight and helpful for fast aggregations with out launching a heavier software like Python or R, which makes it best for shell-based ETL or exploratory evaluation. However it’s not designed for very giant datasets or advanced analytics, the place specialised instruments carry out higher. Additionally, grouping very excessive cardinalities could require substantial reminiscence.
# 8. htop
htop is an interactive system monitor and course of viewer that gives stay insights into CPU, reminiscence, and I/O utilization per course of. When operating heavy pipelines or mannequin coaching, htop is extraordinarily helpful for monitoring useful resource consumption and figuring out bottlenecks. It’s extra user-friendly than conventional high, however being interactive means it doesn’t match properly into automated scripts. It could even be lacking on minimal server setups, and it doesn’t change specialised efficiency instruments (profilers, metrics dashboards).
# 9. git
git is a distributed model management system important for monitoring adjustments to code, scripts, and small information property. For reproducibility, collaboration, branching experiments, and rollback, git is the usual. It integrates with deployment pipelines, CI/CD instruments, and notebooks. Its downside is that it’s not meant for versioning giant binary information, for which Git LFS, DVC, or specialised techniques are higher suited. The branching and merging workflow additionally comes with a studying curve.
# 10. tmux / display screen
Terminal multiplexers like tmux and display screen allow you to run a number of terminal classes in a single window, detach and reattach classes, and resume work after an SSH disconnect. They’re important if it’s worthwhile to run lengthy experiments or pipelines remotely. Whereas tmux is really helpful as a result of its energetic improvement and adaptability, its config and keybindings may be difficult for newcomers, and minimal environments could not have it put in by default.
# Wrapping Up
If you happen to’re getting began, I’d suggest mastering the “core 4”: curl, jq, awk/sed, and git. These are used in all places. Over time, you’ll uncover domain-specific CLIs like SQL shoppers, the DuckDB CLI, or Datasette to fit into your workflow. For additional studying, try the next assets:
- Information Science on the Command Line by Jeroen Janssens
- The Artwork of Command Line on GitHub
- Mark Pearl’s Bash Cheatsheet
- Communities just like the unix & command-line subreddits typically floor helpful tips and new instruments that may develop your toolbox over time.
Kanwal Mehreen is a machine studying engineer and a technical author with a profound ardour for information science and the intersection of AI with drugs. She co-authored the e book “Maximizing Productiveness with ChatGPT”. As a Google Era Scholar 2022 for APAC, she champions range and educational excellence. She’s additionally acknowledged as a Teradata Range in Tech Scholar, Mitacs Globalink Analysis Scholar, and Harvard WeCode Scholar. Kanwal is an ardent advocate for change, having based FEMCodes to empower girls in STEM fields.
