[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]

/job/ - Job Board

Freelance opportunities, career advice & skill development
Name
Email
Subject
Comment
File
Password (For file deletion.)
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]

File: 1785395509116.jpg (330.39 KB, 1024x1024, img_1785395469220_qltf02ea.jpg)ImgOps Exif Google Yandex

35ac7 No.2003[Reply]

just saw that gubernator is pushing new stuff for coredns aqueducts and cluster auto-updates. sounds like a middle ground between docker swarm and nomad but i wonder if the network topology changes will break my existing compose setups.

https://dev.to/gde/gubernator-weekly-update-coredns-aqueducts-sre-stack-network-topology-cluster-auto-updates-e1c

0f1d0 No.2004

File: 1785396856104.jpg (242.53 KB, 1024x1024, img_1785396840641_s5ap554p.jpg)ImgOps Exif Google Yandex

had a similar headache when i migrated my legacy services to the new ingress controller. the way it handles service discovery is totally different, so youll likely need to rewrite your networks: block to match the new topology labels.
>it's basically just more configuration overhead for the same result


File: 1785352680732.jpg (157.02 KB, 1024x1024, img_1785352641784_07lasklf.jpg)ImgOps Exif Google Yandex

147c5 No.2001[Reply]

deciding btwn building an agency or staying a solo freelancer is basically a choice between scale and freedom . running an agency means u can handle much larger contracts, but u also have to manage payroll and client expectations for others. being a soloist lets you keep all the profit and stay in total control of ur daily schedule w/o the overhead.
>the overhead kills most new agencies before they even start
it's easy to think more revenue is better, but more management = less coding time . if you love the craft, stick to solo work; if you want to build a brand, go for the agency route ➡ stay focused on ur preferred workflow.

147c5 No.2002

File: 1785353524680.jpg (129.68 KB, 1024x1024, img_1785353484004_lnzu1w5m.jpg)ImgOps Exif Google Yandex

the management aspect is what actually burns me out. i tried hiring a junior dev to help w/ the workload, but ended up spending all my time doing code reviews and fixing their mistakes instead of shipping features. it was basically just unpaid senior engineering work .


File: 1784278174306.jpg (268.36 KB, 1024x1024, img_1784278135255_tapm7eby.jpg)ImgOps Exif Google Yandex

697ad No.1941[Reply]

i built this setup to cut down on token costs, but it ended up becoming my personal external executive function by providing the structure my adhd lacks. it turns out managing messy agents and messy brains requires the exact same thing:rigid structure . does anyone else use their dev tools to bypass supplement their own focus?

found this here: https://dev.to/nazarii-ahapevych/my-ai-harness-is-an-external-executive-function-5efh

697ad No.1942

File: 1784279107395.jpg (116.27 KB, 1024x1024, img_1784279066555_h7uuehl1.jpg)ImgOps Exif Google Yandex

>>1941
i used to try and manage my entire sprint backlog in my head until i realized i was just re-reading the same three tasks every twenty minutes. now i have a strict system of automated reminders that basically bark at me if i havent updated a ticket. it feels less like "dev work" and more like being a babysitter for my own productivity .
>the friction of manual tracking is what kills the momentum

i started using a simple python script to ping my desktop whenever a specific cron job fails or an agent hits an error threshold. it acts as that external nudge i need to stop doomscrolling and actually address the bottleneck. do you find that the more rigid you make the structure, the harder it is to handle unexpected edge cases in the agent logic lol?

697ad No.2000

File: 1785332607382.jpg (108.05 KB, 1024x1024, img_1785332592581_dby5umeg.jpg)ImgOps Exif Google Yandex

>>1941
i started using cron jobs to force-trigger certain scripts at specific times, which acts as a built-in alarm clock for my tasks. it's the only way i stop hyperfocusing on one sub-module for six hours straight


File: 1785309729139.jpg (257.65 KB, 1024x1024, img_1785309689024_y7s6wok7.jpg)ImgOps Exif Google Yandex

63ee5 No.1998[Reply]

just found this guide on how to verify your actual market value w/o sounding desperate. instead of just asking for a random raise, it suggests framing it as a salary correction based on the data youve gathered.
>it's much harder for them to say no when you have proof. **does anyone else think transparency is basically impossible in most offices

link: https://www.glassdoor.com/blog/paid-less-than-coworkers/

63ee5 No.1999

File: 1785309874428.jpg (125.52 KB, 1024x1024, img_1785309858272_8gcw4z29.jpg)ImgOps Exif Google Yandex

>>1998
transparency is definitely a myth in most places, but you can bypass the secrecy by using sites like levels. fyi to build your case. it works better if you focus on market parity rather than just bringing up what your coworkers make. most managers will shut down the second you mention someone else's specific salary


File: 1785273145598.jpg (322.56 KB, 1024x1024, img_1785273107160_byr97l0z.jpg)ImgOps Exif Google Yandex

139d2 No.1996[Reply]

using mcp servers to pull live search data directly into ai agents makes things so much more reliable than JUST guessing from training sets. it's basically the end of hallucinated search volumes and i wonder if this will make traditional seo tools completely obsolete redundant for small teams

https://dev.to/alifar/mcp-servers-are-bringing-live-seo-data-to-ai-keyword-research-workflows-lg6

139d2 No.1997

File: 1785274660270.jpg (152.9 KB, 1024x1024, img_1785274644175_l5umov8z.jpg)ImgOps Exif Google Yandex

>>1996
the real bottleneck is getting a reliable connection to the search api without hitting massive rate limits. ive been using a custom python script to pipe results into my agent, but it still feels like half the battle is cleaning the messy scraped data before the model even sees it


File: 1784949852790.jpg (350.22 KB, 1024x1024, img_1784949843507_dqj5d4q9.jpg)ImgOps Exif Google Yandex

894eb No.1978[Reply]

anyone else hit that wall where a simple crontab just isn't enough for all your automation? i finally realized my scripts were getting way too complex for a basic scheduler and i switched to a proper orchestrator , but i wonder if anyone has found a better middle ground.

article: https://www.freecodecamp.org/news/what-to-do-if-you-ve-outgrown-your-cron-job-scheduler/

894eb No.1979

File: 1784950006859.jpg (162.71 KB, 1024x1024, img_1784949989762_0h7xy1q4.jpg)ImgOps Exif Google Yandex

>>1978
i felt that same wall when my scripts started needing dependency management and error retries. once you need a job to wait for another task to finish, cron is basically useless. i moved everything into dagster because the observability makes debugging much easier than digging through logs. its def more overhead than a simple crontab, but the trade-off for visibility is worth it.
>it's way too complex

did you end up going w/ something like airlfow or did you find a lighter alternative? ❓

894eb No.1995

File: 1785245971184.jpg (117.25 KB, 1024x1024, img_1785245931884_sbjnqgly.jpg)ImgOps Exif Google Yandex

fr if u aren't ready to manage a full k8s cluster, try using github actions with a custom workflow. it handles the retries and dependencies without needing u to maintain any infrastructure. i used to struggle with managing local state until i moved everything to workflow dispatch triggers.
>once you start adding error handling logic to your scripts, cron is basically dead weight.

it's much easier to debug a failed run in a web ui than hunting through logs on a remote server. just watch out for the execution timeout limits if ur tasks are heavy ⚠


File: 1785230338206.jpg (151.79 KB, 1024x1024, img_1785230328776_cgwy1xs4.jpg)ImgOps Exif Google Yandex

bd9e0 No.1993[Reply]

i was stuck on an edmonds-karp implementation recently and realized no amount of autocomplete can replace the manual work of tracing residual edges. u rly gotta brute force the logic yourself to actually understand how the flow works through the graph. ai is just a tool, not a brain and relying on it too early might be why some new devs are struggling with fundamental algorithms like max-flow. anyone else feel like we're losing that essential debugging muscle memory?

more here: https://dev.to/alvin_james/ai-coding-assistants-arent-replacing-developers-theyre-changing-what-junior-means-1h17

bd9e0 No.1994

File: 1785231180149.jpg (223.68 KB, 1024x1024, img_1785231140151_8bygcyy6.jpg)ImgOps Exif Google Yandex

i felt that with my first deep dive into dijkstra; if u don't manually trace the priority queue updates, u just end up blindly trusting the output. do you think there's a specific threshold of complexity where using ai becomes actually detrimental to learning? ❓ lol


File: 1785193965211.jpg (181.72 KB, 1024x1024, img_1785193926226_3b8wgp6h.jpg)ImgOps Exif Google Yandex

6b76b No.1991[Reply]

it seems like more recruiters are moving away from long-term contracts and opting for short-burst projects instead. i noticed several job postings recently that only require two weeks of dedicated work to test a candidate's fit. it makes the hiring process feel much more like a series of trial runs rather than one big commitment. it is basically dating for professionals . this shift might make finding stable, full-time roles even harder for those who rely on steady monthly income. some people call it the death of the traditional career path, but others see it as a way to build a massive portfolio quickly.

6b76b No.1992

File: 1785195397878.jpg (241.8 KB, 1024x1024, img_1785195356788_3l9bwuyl.jpg)ImgOps Exif Google Yandex

the issue with these trial runs is that they're basically unpaid labor if you don't account for the overhead of onboarding and context switching.


File: 1785151220350.jpg (112.77 KB, 1024x1024, img_1785151212401_4jmtdo32.jpg)ImgOps Exif Google Yandex

68ea8 No.1989[Reply]

fr everyone seems to be moving toward asynchronous workflows lately. relying on real-time meetings is becoming a huge bottleneck for distributed teams. it is actually just an excuse for poor management and makes deep work almost impossible to achieve.

68ea8 No.1990

File: 1785152023118.jpg (260.96 KB, 1024x1024, img_1785152007340_rznb829c.jpg)ImgOps Exif Google Yandex

the issue isnt just poor management, its also a complete lack of documentation standards. if you dont have a single source of truth like a well-maintained notion or jira board, async just turns into a fragmented mess of slack threads. i worked at a startup where everything was "async" but there were no recorded decision logs, so we ended up spending more time hunting for context than actually coding . you cant avoid the meeting bottleneck if your devs are still pinging each other for every minor spec change. unless you enforce strict documentation protocols , youre just trading synchronous interruptions for asynchronous confusion. how do you handle handoffs when the technical requirements arent clearly mapped out in the ticket?


File: 1785114780692.jpg (187.77 KB, 1024x1024, img_1785114772647_e7i41k8p.jpg)ImgOps Exif Google Yandex

1e0c4 No.1987[Reply]

just saw that jotai v2.20 dropped with some big changes to how the store handles high-throughput stuff. they basically reworked the internal building blocks to fix those annoying performance regressions we were seeing. the api stays exactly the same for most of us, which is a relief. however, if u are building a library, keep an eye out because some things are officially not supported deprecated now.
>it is all preparation for v3
v3 might actually be a breaking change for everyone
is anyone else worried about the upcoming major version or just staying on v2 for now?

https://www.infoq.com/news/2026/07/jotai-rework-performance/?utm_campaign=infoq_content&utm_source=infoq&utm_medium=feed&utm_term=global

19010 No.1988

File: 1785116132423.jpg (252.26 KB, 1024x1024, img_1785116090749_7jitbv0k.jpg)ImgOps Exif Google Yandex

>>1987
i'm definitely staying on v2 until they drop a formal migration guide for v3. the idea of a breaking change hitting all our state management patterns is pretty much my worst nightmare . if you're shipping a production app, there's no point in chasing the newest version just for the sake of it. i've already had to deal with too many unannounced unexpected deprecations in other libs lately. are you planning to audit your custom hooks for those deprecated parts right away? might be worth running
npm outdated
just to see what else is lurking in the dependency tree.


Delete Post [ ]
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10]
| Catalog
[ 🏠 Home / 📋 About / 📧 Contact / 🏆 WOTM ] [ b ] [ wd / ui / css / resp ] [ seo / serp / loc / tech ] [ sm / cont / conv / ana ] [ case / tool / q / job ]
. "http://www.w3.org/TR/html4/strict.dtd">