Hiring Software Developers The New Way.

2026 Day 176. #BusinessDays.

While software engineering is not going away soon (we need someone to develop the models), we must do away with the traditional hiring systems. The workflow has changed extremely yet hiring is still the same.

I would chalk up the lack of change to corporate bureaucracy, HR incompetence and the focus on removing software engineers from the system entirely. I have worked with a large team, with non-technical clients, small freelancing teams, and other individual developers. This is how I expect the new wave of hiring to look like.

The major shift needs to happen from syntax to thoughts and understanding. Every interview needs only two parts.

1. Good Conversation. Not your everyday small-talk but should instead focus on gaining an understanding of the various technologies the person has worked with how can they align with the needs of the business.
2. Short Writing Prompt. Ask the developers to imagine writing a long prompt to solve a certain problem. They must assume that the LLM will build the entire thing in one shot and must provide the best prompt possible.

Why will this work?

As a new age developer you do not need to prove you can write code. The majority of the boilerplate code is written for you. The task is putting a greater emphasis on software engineering over simply writing code and praying.

Markdown files with relevant skills will set up the LLM to be smarter than most developers out there very quickly.

Coding standards can be adhered to better than most humans would stick to. There will be less friction because you do not need to consider the relationship with your (digital) co-worker. You can be blunt about the cons and get quick rewrites.

Running AI is expensive. The costs are not being fully transferred to the customer just yet. Here is the google search result for predatory pricing. I will let you wonder how it applies to the situation.

A deeper look at Good Conversation.

This would be a great test of thinking on your feet. As an AI-aided developer the majority of my problems are not syntactical. It is based off of visual issues, convoluted issues that the result of several smaller ones coming together, and the problems faced in working/integrating third party frameworks/software.

A good developers need to have excellent reading and writing skills. To be able to skim through all of the plan that an LLM generates, look to correct it, make improvements, these are all the new job descriptions in the field of software engineering. Being able to hold a conversation with a human over such discourse is a great test of the quick thinking abilities.

This is the test of the comprehension and investigative skills of a developers. During the conversation the developers must observe a problem, present possible reasons for the issue, decide how to eliminate the issues and arrive at the root-cause.

The problem could be just code, a mix of LLM output and code, or lot larger context depending on the seniority of the role that the developer is looking at.

A deeper look at Short Writing Prompt.

Good prompts = cheaper, better results. Any developer who has used LLMs will tell you this is true. If prompting is the majority of the job and good prompts will save the engineer time, the company money, then this essentially becomes the most important skill.

Testing for this is also relatively simple. The candidate writes a prompt (either live or prepared beforehand) and then the interviewer questions the decisions based on the prompt they have provided.

The challenge would be to have people competent enough to develop scenarios that can be one-shotted effectively in relatively short prompts but only with accurate key points. If a person is competent enough to do this, then they would also make a great interviewer.

People will tend to overdo this and make it overly complex but that is the fault in our stars (humans).

While these might not be the end all, it is certainly a step in the right direction. The quality of the conversation will be dependent on the skill level the interviewee is at. There could be more than one prompt across multiple interviews to test multiple areas individually, depends entirely on the specifics of the job.

The biggest problem in hiring always has been the human loop. We need humans (imperfect) to judge other humans (imperfect) in very limited encounters.

What do you imagine could be the problems associated with this new format of job interviews? Could this work in alternate fields as well?

See you tomorrow.