Toni: So, going back to the job application process, if companies are struggling with this level of complexity as far as notifying me when my order has been delivered, can we fully trust these complicated HR processes to be running successfully?
Darren Arney: Well, that I think is where a lot of the actual work is going to be in automation. And, you know, hopefully it's something I can help find myself. But you just said it right. The trick isn't that nobody is resisting automation. People are resisting the bad automation they already have.
Toni: A resume turns years of work into a handful of bullet points. It shows what someone did, but it doesn't highlight how they think. I'm Toni Holcombe and you're listening to Tell Me About a Time: The Stories Behind the Resumes. Real people, real stories. Let's get them hired.
Toni: All right. Well, welcome. I have Darren Arney with me today. Darren, hey.
Darren Arney: Nice to be here, Toni.
Toni: Thank you for joining. I appreciate it. Why don't you get us started and tell us a little bit about your job history? Tell us about your experience.
Darren Arney: Sure. Yeah. So, I mostly have been working in automation for the past 14 years, and that's kind of a very broad statement that has a lot of specific stops on the way. Most relevant, starting about 15 years or so ago, I worked at a place up here in Indianapolis, up on the northwest side. Did a lot of agricultural technology in a lab there that did some genetic sort of DNA extraction from plant tissue, and I went from being a contractor in that lab to being a full-time employee. Part of the deal there was that I worked from 11:00 a.m. to 8:00 p.m. So, that sort of allowed me to indulge my tendency to just want to figure things out and tinker with stuff a bit, in an environment where it's moderately safe to do it. I can't, you know, break the work we're working on, but it's certainly an environment that has the right kind of stakes to be able to experiment, and make sure that you're providing and not just experimenting for the sake of doing it, that you're building something that lets you improve the process you're already working on. I started from there, and then I left there in 2017. I had really gotten into what I was doing and enjoyed the impact that I was able to have, but I was sort of feeling like I was doing that for one set of scientific assays for one specific thing, but the knowledge and the tools that I was learning allowed me to have a lot larger impact. And then after that I went to a place called Biosero, that's based out of San Diego, where I was for the past 9 years or so. And there, the idea that I had was like, well, I really like writing these methods and building out all these automation protocols for these scientific workflows, and if I could just do that for everybody, for all the kinds of things that they need, that'd be a pretty great job. And I really enjoyed the time that I was able to spend there and do that. You know, I think with any job, especially when it's one that you really like, there can be some moments where it's a be careful what you wish for a bit, but at the same time, that just keeps the kinds of challenges that keep stuff from being easy. So, without being too specific, I've been able to help improve a lot of processes, create a lot of impact by automating a lot of processes, but you can get a bit meta when you talk about automating the automation. And at the center of it all, I just like tinkering with a lot of things and making sure that I'm making them better when I do so. That's kind of been the through line throughout the past 15 years or so. And I think recently, with the adoption of a lot of agentic coding tools and AI more on the technical side of things, there's been an incredible explosion of opportunity to take the principles that I've learned with automation and apply those to kind of anything, really. A lot of what brings success in automation is making sure that you're willing to fail a lot so that you can figure out the one way that works, especially when dealing with physical automation. There's a lot of stuff that you're just not going to be able to tell until you see it physically in motion, and that teaches you a lot of different kinds of experimentation. You know, you can't just code it. That's not going to be the solution. You have to code it right. You have to make sure that it does the job that it's supposed to, which in my previous experience was a scientific test for, say, doing a COVID test or a different genetic marker in a plant or a cell line or something.
Toni: Yeah. So when you talk about improving processes and doing the automation, what kinds of processes, and feel free to generalize if you need to, because I'm sure that's proprietary information, what kind of things were you automating?
Darren Arney: Yeah. So, I think this is the tricky part now with the current era of AI and stuff that we're in. Everybody's willing to automate your basic day-to-day tasks, right? Like I opened Instagram yesterday and it's saying, "Hey, do you want me to book a dinner reservation for you?" And absolutely not. I am capable of doing that. I can operate my own phone, you know. But again, 15 years ago, or say a hundred years ago before we had the assembly line, there was a person that had to do a little simple task every single time. And now we're so full of automation that we kind of take that for granted. But it's easy to generalize. Again, when you say doing a COVID test, right, that means you started with a nose swab and a test tube and then some solution in there that is going to allow you to turn that into a web page that you open that says I have COVID or I don't have COVID. Now, of course, you can do it on a card that you can buy at the drugstore, but in 2020, for example, you needed to be able to automate the science for that in a reproducible fashion that was reliable, that ideally didn't require a lot of people, both for human-to-human contact concerns in the lab, but also the sheer volume of tests that they were working on. And a few of the different projects that I worked on were helping to automate those steps, right? So, for example, one of the steps after taking the tube full of the nose swab would be to take a solution out of that and move it into a tube, and then take that solution and move part of it into a tube and then part of it into a different tube, and then add solution to those two tubes. And you had to have a person doing all that, right? And now that is something that you can program on a robot to have it do. But even when you write the perfect code for it, you still have to sit there and watch it and make sure that it's actually picking up a tip, or it looks like it's pulling the right amount in, cuz you want to validate what you're doing before you actually have it dispense into that plate. Now, I didn't do the liquid handling automation directly as much. I did a lot more of the higher order, like delivering the plates to the liquid handlers, but for the purposes of this conversation it's a bit of splitting hairs. And, you know, I have a degree in biotechnology, but it's been over a decade since I've taken a biochemistry class, and I don't think that the thing that allowed me to do the automation was knowing all of the science. But having that background allowed me to make some better judgment calls in different places. And at the end of the day, a lot of automation, and even really a lot of AI stuff, is just providing the human judgment that allows the computer to make those judgment calls better, basically. But it still requires a lot of human input.
Toni: You know, so two things that I want to touch on real quick. The first one being, just by way of experience, I remember taking the COVID test and waiting for the website to tell me whether or not we tested positive. I had never thought about the process that got us from point A to point B in that.
Darren Arney: Exactly. Yeah.
Toni: So that is like really fascinating to me.
Darren Arney: I was just going to say, I was far less frontline on that than a lot of much more important people. But it is some of the work that we did. You know, I don't want you to think I was in there doing the swabs myself. I was getting them, more often than not, to go do the work in the labs. But yeah, continue.
Toni: I just think it's really super fascinating. And then the other thing, with the human judgment. The position that I have kind of taken is that we shouldn't call it human in the loop. We should call it like AI in the loop, because the judgment still has to come from the human.
Darren Arney: I think it's refocusing on what the loop is. If the loop is the mechanical work, it was probably an easier way to view it at the earlier stages, and I say that being, let's say conservatively, less than a year ago, where people were talking about human in the loop operations. But the way that I think about it is, basically, I'm the human and I send the robot to go off in a loop and do all this work, and then it comes back to me and says, "Hey boss, did I do good? Look at all this stuff." And then I say you did or you didn't. And either way is fine. But remember why this was good and why this was bad. And eventually it should remember why you made those calls and then keep making them. And the next time it'll say, you know, because you said this last time, I made this judgment call that way. And that's where it does become a lot like conventional automation, because that's really tedious, but it's also the stuff you need to pay attention to, right? To make sure that you can have trust in those systems. Cuz I think obviously trust is the biggest issue that we have with adoption in any technology, and this is no different. The problem is, I think with AI, a lot of people have to build that trust themselves and don't want to, because why would they have to? They've never had to do that with their phone before, you know.
Toni: Right. Okay. Tell me about a time where you made a significant impact on an organization, and what that looked like, what that process looked like, and what the outcome was.
Darren Arney: Sure. I've talked a lot before about a similar technology, right? We built a system to do cell line development, which is to say that when they make any biologic drug, they have to make sure that, and I'm probably butchering this horribly, but let's say all of the Zepbound going into one pen has to have provenance down to the original cell that split into two cells that split into four cells that created the cell line. So then they know all of the Zepbound being produced by these cells is the exact same Zepbound, right? All the generating cells are as identical as they can be. But you can use that technology to create Zepbound, or you can use it to create COVID antibodies. The weight loss drugs were the focus of why we built that system in the first place, but it was also for any number of things. And when they needed it, they could utilize it, because we built it robustly enough to build the COVID antibody treatments for Lilly. So, it's one small tooth in one gear in a giant machine, but I feel like I get to own that tooth at least, and that feels pretty nice. And the best part about that for me was that I didn't really have to do anything beyond what I already did. Now granted, that's not to undermine a lot of the important work done by the people that were loading and unloading the system. I just programmed it. All of the good data that comes out of it is because somebody's working on the program that I did, and it's not just pressing a button and working by itself forever, right?
Toni: I am going to program all of my automations in my loops to say, "Hey boss, did I do a good job?" I mean, it's just for fun.
Darren Arney: Well, it works, right? Your brain is evolutionarily wired to work a certain way, and primarily that's with communicating with people. We're social animals and we want to talk to people. And it can get a little dangerous when you start making the computer people. Obviously, there's a whole subgenre of people that go a bit too deep in how they talk to AI, I think. But I don't think it is inaccurate. I talk to my car like it's a person, especially when it's not working well, you know, and it's just an extension of that. Your car can't talk back most of the time, or it's not using words. That gets into a lot of the weirder stuff, I think. But if you can just view it as a mirror most of the time, that's really the example that you want to have, cuz what you're doing there is saying, you know, how's it going, are we having a good day, is everything going fine, and you're really just saying that to yourself. But then that starts sounding weirder, so you just keep going that way.
Toni: I think I give myself the best advice sometimes. I'm just kidding.
Darren Arney: Listen, it's important to have an internal dialogue. Externalizing that can be fine too.
Toni: So, what you are saying about the COVID testing process: was that process that you built out something that became repeatable for other processes? So we could take it out of the COVID part and put it into something else?
Darren Arney: I think the way that it worked was the inverse of that. We had built a system that did a good job of the stage it was responsible for in the early cell line development. So it happened to be a coincidence that the process that we had automated was what they also needed for some of the early stage COVID antibody development, and they already had a system that could do it, so they leveraged what they had in place. I think the kind of thing that you're pointing at is a parallel but differently important problem, which is that what AI is really great at is reproducing a lot of the mechanical operations that you need to have happen when you do this stuff, right? Cuz when I talk about doing that COVID test, to keep myself out of the weeds, we'll just say something needs to go from A to B, then inside B something needs to happen. Now, any number of things inside B can be true, and those would still qualify as valid science for a step in a COVID protocol or for a DNA extraction test. But even those steps have subcomponents, and systems exist to let you not realize things are just copying and pasting in the background a lot. But that's what AI is really good at. And a lot of the machine learning stuff that we need is to say, okay, I have these 45 steps that are actually required. I don't want to have to rewrite these 45 steps every time. That's exactly what AI is great at, right? And if you can find ways to leverage that sort of functionality in different systems, then that is exactly what you want. Because I had to build a complex series of liquid handling operations to make the cell line development system work, and those complex operations would be useful in a different system. But the nature of most programming is you can rarely just copy and paste from one environment to another. There's a whole Byzantine series of things you have to do, and most of the time you just end up rewriting something from scratch. Nobody ever has to rewrite anything from scratch anymore, I think, is kind of the lesson. And even if it's that bare minimum of effort saving, then you know, never having to say "I can't believe I have to do this again" is a huge relief for somebody who's had to do that a lot, and I think that applies to most jobs, right?
Toni: So I have two questions, and I'll introduce them one at a time so you can answer them. The first: what got you interested in biochemistry?
Darren Arney: I just had most of a degree from Purdue and I needed to finish up the rest of it. I started in engineering at Purdue and ended up with about 80% of a food science degree. I was bartending more than I was going to class for the last couple years of that. So then I came back and was working jobs at night and going to Ivy Tech during the day. I do truly enjoy all of the hard bench science, and biology especially, because it kind of exists in that bridge between chemistry and physics that also keeps everybody alive. So it's nice. But I had my own problems with being able to finish that. And once I'd gotten a bit older, I was able to get an associate's in biotechnology. And then while I was working at the first place I was talking about, I was able to finish a bachelor's in biotechnology. So it was basically like, I have most of this degree. I can work to finish paying off the couple credits I have to get the associate's and then the bachelor's, cuz I just needed to have something. I was paying down student loans and I was like, I'm not going to continue to pay all these against nothing. So yeah, I think it was a really good introduction to the part that I wouldn't have a good background in if I went into this kind of field on a pure tech side. Cuz the thing I used to say a while ago was that you can't go online and do a biochemistry lab, but you can go online and learn Python. And I guess that's kind of still true, but you don't even have to learn Python anymore. You can just have AI do it for you, for the most part. But that's maybe a good comparison too, right? Cuz you can have AI write the Python for you, but if you can't read it a little bit, the likelihood that you're going to let something bad slip through is way higher.
Toni: Right. Right.
Darren Arney: And yeah, I don't know. Does that make sense? Sorry.
Toni: Yeah, that makes sense. And it's totally okay. I didn't go to college until I was in my 30s. I tried right out of high school, but I was like, I'm going to go experience life, and then I had enough life and I was like, I need to go back to school. So I get it. And the second question I had is, when we think about AI, how is traditional RPA different from the tools that are available today?
Darren Arney: Yeah. So I think that it's interesting, cuz I have heard of RPA as a term, and I understand that it means robotic process automation, and I've also come to understand, within the realm of looking for jobs and stuff now, that it means a completely different thing than my level of experience. Which is interesting, cuz the concepts are functionally similar. But like, if I want to apply for a job with 5 years of DeltaV experience, I don't have that, but I could probably learn something like that if I needed to, because I've already learned probably more complex systems. But I think you get into a sort of industry-specific credentialing there that becomes a bit more difficult the longer you're not part of a specific industry, maybe. Does that make sense?
Toni: Yeah, that makes sense.
Darren Arney: So yeah, I have a lot of broad automation knowledge and can apply those concepts pretty reliably across any number of domains. But if I wanted to apply for a job where I was programming PLCs in a factory line, completely different skill set.
Toni: Yeah. Completely different, because then you have software automation versus real life physical automation stuff.
Darren Arney: Yeah. And for the RPA stuff, is that mostly with regards to website automations or different back end stuff?
Toni: I was thinking of it more like in a previous role, where they called RPA basically going into software and scraping information and data and then putting it over here.
Darren Arney: I see. Yeah, that's certainly different. That's something that I don't have individual experience with before. I've been having Claude do a bunch of it for me, basically.
Toni: Okay. So, how did you find yourself in a position where you're looking for a new role?
Darren Arney: Well, like I was saying, the last time I left my previous employer and went to Biosero, I was looking for probably about 10 months, in that weird, I think we tried to call it quiet quitting a few years ago. But I think generally that just meant looking for a job while you have a job, which is what everybody should do. Maybe not a great thing to say on a job seeking podcast, but I clearly made the opposite decision, so I'm happy to stand by it. And basically, I was really lucky to land the job at Biosero that I did when I did. I had a really great mentor there that helped me understand how a smaller business like that works, and I got a really great opportunity to have a lot more responsibility there over a number of roles across a number of years. I think what ended up happening, from where I sit, is a lot of the same stuff that'll happen after a company gets acquired, a couple years down the road. There's not really one major cause or concern, but it doesn't really feel like the place that it was when you were there. I think that's probably the most worth saying about it. And beyond that, there's been a lot of opportunity with AI-assisted software. I worked in automation software, which is about as parallel to AI as you can get before it started popping off. And so just in my own time, I've been building a lot of internal tools and automations for my own house, and I started to understand the value that this stuff has, cuz it allows me to take all the stuff that I used to bore a lot of people with and say, okay, here's how I could do that for you. And that's not exactly as easy an operation as that makes it sound, cuz trying to find work and look for jobs and do all the things you need to do to put food on the table are still important and challenging and time bound and everything else that they need to be. But this time, it seemed like it was useful to try and go start doing my own thing. And so I left at the end of April.
Toni: Okay.
Darren Arney: And I've been trying to do some consulting work and, I'd say, ad hoc software development for other people. But like I said, being an entrepreneur was not really in my wheelhouse, and I think I probably leaned into Claude a little bit too much, trying to just say, oh, I can automate this too. But you've got to be careful with that.
Toni: Okay. So now you're looking for a new role, kind of, sort of, but you're also building out your own AI consultancy, process automation practice. So what are some of the things that you have done in your job search? What are some of the techniques you've adopted?
Darren Arney: This is probably not the most hygienic approach, but I really have just been using AI to rewrite resumes appropriately for every role that I'm applying to, and then reviewing them to make sure they're honest, and submitting them through whatever portals you need. Sometimes they'll have one where you just upload one for the platform and can just kind of batch apply whenever you apply for something. But I think any sort of advice around the normal platforms and stuff is probably not worth the amount of time it would take to listen to it. I think I have gotten some good contract work off of just Indeed. But I couldn't have told you why. I have applied for hundreds, easily, of similar roles, and there's no rhyme or reason as to why that one took versus the others. One thing that you can learn in automation is, if you keep doing something and it keeps making your life harder, you should probably find a way to stop doing it. And I don't think just throwing a million resumes into the wind and sitting here and waiting is going to really make me feel any better about it, you know, because it doesn't come back with hundreds of other jobs and stuff. So I think networking like this, and doing other kinds of things within the industry that you're in, is probably much more valuable than any sort of blank application process. But I think it's also just gruesomely random, too.
Toni: Yeah, it is. It feels very random.
Darren Arney: Yeah. And it is. It makes sense. I was the manager of my team at my previous job for a couple of years, and I had to try and hold on to a lot of people during some times when it was really tough, and I got to hire a number of people. And even then, you're hiring because you're busy, and you're busy because you don't have time to do the work you need to do. So if on top of that you get 500 resumes fire-hosed in from Indeed, you're not reading those. And even if you have somebody in HR that's supposed to read those, they're not reading those. I know there's a lot of scanners and stuff for them now, but that was always true. I remember hearing about putting all the keywords in a one-point font in white at the bottom of your resume, that trick, from 15, 20 years ago. It's not new stuff, trying to trick a scanner. But searching for a job is supposed to be a representative thing of here's who I am and here's how I can help you. That's not part of it, you know? It just doesn't feel like I'm being genuine if I'm doing that. And it's just more time you've got to waste when you could be making yourself better, or learning a new skill or process, or learning that next programming language, I guess. At least that's more relevant to my experience.
Toni: Yeah, I agree. And I think companies aren't all that great yet at automations. Like, I order this collagen peptide for my coffee in the morning, and I will order it and a couple hours later I will get an email from the company saying, "Oh, were you still interested in this product?" And I'm like, "I bought it. What are you talking about?" And then it will get delivered the next day and I'll get three emails. I'll get an email that says it's been shipped, an email that says out for delivery, and an email that it has been delivered, but it's not always in that order, and they always land within like 10 minutes of each other.
Darren Arney: Oh, yeah. See, and you don't need those. I think that's a big problem that nobody has a solution to yet, cuz really what AI should kill, if it's this omnipotent thing, is 75 notification emails. But if I was the person building that platform, I would say this is great, now you can see when this thing is done. But that's so you know the software isn't broken. That's not because you care about the package actually being there. I think it's the outcomes of a lot of those automations that aren't really considered, and that's why you have, again, the Instagram bot saying, "Hey, do you want me to get you a dinner reservation?" No, I want you to show me pictures of my friends and family. That's what you're supposed to do.
Toni: Yeah. So, going back to the job application process, if companies are struggling with this level of complexity as far as notifying me when my order has been delivered, can we fully trust these complicated HR processes to be running successfully?
Darren Arney: Well, that I think is where a lot of the actual work is going to be in automation. And, you know, hopefully it's something I can help find myself. But you just said it right. The trick isn't that nobody is resisting automation. People are resisting the bad automation they already have.
Toni: Yes.
Darren Arney: It's bespoke. Cuz, you know, your dentist and your doctor likely have the same software package, or they could theoretically, that deploys your appointment reminders, and they all kind of look the same. And one sends them, for some reason, a week before your appointment, and one sends them a day before your appointment. And depending on who you are and how you live your life, maybe both are useful. I have to keep a calendar for everything, otherwise I'll forget. So those are useful to remind me that things are in my calendar, but otherwise that's a quality check on my process. There's a million ways you could fix that, but all of those are going to be bespoke to the person that actually cares about when they need to be reminded about their appointment. And you've shown them with software that you could build infinite menus to say, here's how I want to be reminded about my appointment, or when my package is going to show up, but nobody wants to spend the time configuring those either, right? So, it's a question I don't have the answer to, because I think we just tried to mine everybody's data until we stopped letting them, and assumed the answer would be in there. But I don't think it is, otherwise we wouldn't still get those emails.
Toni: Right. Right.
Darren Arney: There's not any other way to broadly apply that without saying, okay, well, we need to know when Toni opened her phone, or the last time she opened her email, to see whether or not we should remind her, or whether she opened the website she ordered from, to see if we should surface this in the email. But that's a huge privacy concern, right? It is something, theoretically, I guess you could build yourself, but that's a bit of navel gazing, I think, more than functional utility, probably.
Toni: Right. I feel like this is also another side quest in my brain. I feel like eventually we'll move away, or at least I hope we move away, from all the apps, and we'll just kind of have more of a universal interface with all these different products and data. I think it'll just simplify, because everybody wants you to either scan their QR code, which I don't always trust, or download their app. And I'm not downloading an app to order a smoothie. I'm sorry.
Darren Arney: Well, that's the pressure though, right? I think it's interesting, because we don't have a solution that isn't that, because you can't consolidate it entirely without giving all of that control to one primary vendor, I think. And the closest we have now is, again, having say Apple or Google as where your phone lives. But neither of them can consolidate all of that without also owning the data on all of that, which is the tricky part that somebody way smarter than me is probably going to figure out eventually. But, and I would say this is complete conjecture, it seems like OpenAI is trying to go more into that. I saw they announced some marketplace thing recently. But then does that just mean you're buying your apps on the OpenAI marketplace and there's a little AI button on it now? I don't know what that means yet, right? It's like we all want Star Trek, but then somebody's got to be the one company in charge of it. I don't think anybody wants one company doing that yet, you know.
Toni: No, I agree. So, tell me about your ideal role. What are you looking for next?
Darren Arney: Sure. So I have talked to a couple of people that I used to work with in the previous life sciences automation industry, and I think the type of thing that I would be really good at, and the thing that I've had some experience with in my personal development in the recent past, is building a lot of knowledge based systems for internal automations. So, kind of like we've just been talking about, right? There's all of these individual bespoke solutions that would actually solve a lot of the problems we have. But on the personal level, individuals are fickle enough that we may not want to change the scheduling of our emails as frequently as we change our mind about how often we want those things, right? But an organizational process or a business process, those are defined things that you need to do. So I think building a lot of knowledge bases around business processes, keeping those isolated off of the internet so that confidential business information isn't just thrown in the cloud like everything else, that's a comparably inexpensive thing to do. And I think there's going to be a market for that. I just need to figure out where it is. And that would allow me to keep doing what I've been doing, right? Building these systems for people that have a lot of individual business impact for them, but allows me to say, "Oh, great. I learned a little bit about this," you know, that you can migrate from QuickBooks to a self-hosted ERP system. And then I put that in a box for somebody and saved them two grand a year on their QuickBooks subscription. Or for the next thing, I put all of their business processes into this box, so that an office can query what the next invoice operation is supposed to be when they're hiring people or something. Just something that would allow me to keep using those skills and keep building them, and sort of just helping people generally. I think the best part of the long-term vision on my last couple of jobs has been, even when the work's hard, it's going towards something that's helpful to people. And whether that's feeding them, or helping develop medicines for them, or helping diagnostic testing, any kind of stuff, it felt like the work that I was doing was worth it. I'm not a person that likes not being busy. And so I think when you're like that, you want to make sure you're putting your time into things that are worth it, and aren't just things to put your time into, you know. So, I don't know. Does that make sense?
Toni: No, that's perfect. Yeah, absolutely. So thank you very much. I appreciate your time. I hope that you find the most perfect next role. You are a very genuine person, very authentic, and very intelligent, and you deserve all the best things.
Darren Arney: I appreciate that, Toni, and likewise. And yeah, thank you very much for taking the time.
Toni: You're welcome.
