I haven't used DSPy this much, but as I understand it: this lang is more like an orchestration DSL for writing and running LLM conversations and tools, whereas DSPy is a framework that compiles and optimizes LLM programs into better-performing prompts...like DSPy has automatic improvement of pipelines using its compilers/optimizers. With DSPy you deal with modules and signatures.
Nice article talking about the common pitfalls popularized by likes of Google and Palantir a decade or so ago. Some of these methods work if you are hiring enmasse and are ok to tradeoff identifying some outliers (like in homebrew creator's example) to efficiency.
This is one of the reasons I built nitrohire.co to be able to get more relevant information about a candidate in a different way! The basic question is why try to guess something about a candidate that their peers might already know and help you make more informed decisions
Last year recruitment was a pain. This year, it's slightly better, still the process hasn't changed much. After spending an inordinate amount of time interviewing, I brainstormed and launched nitrohire.co last month after a few months of alpha usage. Nitrohire helps hiring managers get quick and useful information about candidates without any effort.
Currently doing on and off in the vicinity of a grand a month.
I built nitrohire.co, a reference checking tool for hiring managers and also did a Show HN recently [1].
The revenues are a few $100s currently though I have not been doing a lot of marketing for it yet since the beta launch resulted in a lot of feedback that I am trying to evaluate and address.
Also, if someone is interested in hiring/recruitment/professional networking space, would love to connect! Email in profile
I am Devang, creator of NitroHire. It's still rough at the edges, yet I am very excited to share this on Show HN!
Many of us have been through the experience of either sitting through interviews or taking interviews, and very quickly the interview process becomes a really big chore of asking the same canned questions, and hearing or giving the same canned responses most of the times.
Having sat through taking hundreds of interviews in the past decade or so, every now and then I would ask that there is got to be a better way to do this. I was bored to death taking skills tests, or asking DS-Algo, product, design questions, and trying to glean insights and read between the lines. And I could see with the candidates also having interviewing multiple places at the same time that they were equally jaded yet had to feign a lot of interest and enthusiasm.
So I thought there should be a better way to tackle this, and one of the thoughts was to get first hand information about these aspects of a candidate from people who already know the candidates, rather than to try to intuit about the candidates. This is how the draft of NitroHire started, to help get this information to a hiring manager fast, so that the subsequent interviews can be shorter, more interesting and fun because of the additional context.
The first version of NitroHire supports getting reference check very quickly with predefined questions templates for almost all roles in product and development teams (build your own questionnaire from templates feature coming soon), takes less than a minute for hiring managers and candidates, and gives a very detailed and info rich report to the hiring manager.
I would love for you to try our NitroHire and also learn from your experiences and feedback on how we can make interviews suck less and make hiring decisions faster!
Looking for a co-founder to try and build different ways to interview and assess candidates. There is an MVP currently that I have developed which has helped garner good feedback. Now it's time to get these users and multiple other perspectives to get to the actual product.
