How to Start Freelancing as an Engineer (Contracts vs Projects)

Author: Dave Ebbelaar · Watch on YouTube

Video authored by the channel shown, not by IANIA; all rights belong to the creator. Transcript and summary generated automatically; the original-language content is preserved.

Original language: English

Original video: How to Start Freelancing as an Engineer (Contracts vs Projects) · Channel: Dave Ebbelaar · 17:45 · Watch on YouTube

Topics: AI agents, Task and workflow automation, Open source and software ecosystems

Summary

The video distinguishes between long-term freelance contracts, typically paid hourly or by the day, and project-based work, where a freelancer delivers a defined outcome for a fixed price. The speaker argues that long-term contracts account for most of the income of established tech freelancers, while solo freelancers can find project opportunities automating routine back-office work for small and medium-sized businesses. He recommends combining a contract with smaller projects and, over time, maintenance retainers, which can provide recurring income and room to build independent software. He also describes how AI coding agents have changed his agency’s capacity and invites viewers to a free freelancing masterclass.

Key points

Long-term contracts resemble full-time tech roles, often run for months, and are usually paid at an hourly or daily rate.
Project-based work has a defined scope and fixed price, but requires more frequent sales, outreach, and proposals.
The speaker says 80% of revenue for long-established, full-time tech freelancers comes from long-term contracts.
Long-term contract roles generally require several years of relevant experience and a demonstrable track record.
Solo freelancers are more suited to end-to-end project work for small and medium-sized businesses than for enterprises.
Common AI project opportunities include automating invoices, emails, document handling, reporting, and customer support.
Clients often care more about business outcomes than the technical frameworks or models used to deliver an automation.
The speaker recommends combining long-term contracts with project work and maintenance retainers.
Recurring maintenance income can give freelancers more flexibility to develop their own software products.
The speaker says AI coding agents have enabled his two-person agency to do more work than its previous six-person team.

Chapters

Start	Chapter	Summary
00:00	Two types of freelance work	The speaker introduces long-term contracts and project-based work, comparing their pay structures and how a freelancer’s year might look.
03:09	Why long-term contracts matter	He explains the sales burden of short projects, argues that long-term contracts make up most established freelancers’ revenue, and outlines the experience clients expect.
06:17	Project work for smaller businesses	The video shifts to solo-freelancer projects, focusing on back-office automation opportunities and the difference between enterprise and small-business clients.
10:26	Delivering automation projects	The speaker describes background automations, on-demand assistants, and examples of projects his company has built for small and medium-sized businesses.
12:30	Combining contracts, projects, and retainers	He recommends layering project work and maintenance retainers onto long-term contracts, and explains how recurring revenue supports building software.
16:09	Freelancing masterclass and closing	The speaker summarizes his suggested freelance model and promotes a free online masterclass about freelancing with AI coding agents.

Quotes

long-term contracts and project-based work — 00:31

80% of the revenue from freelancers in tech that do this full-time for a long time, 80% of that revenue comes from long-term contracts. — 04:12

This is the huge opportunity. — 08:21

Long-term contracts, project-based work on top of that, plus maintenance retainers — 16:09

Concepts and entities

long-term contracts, project-based work, back office automation, maintenance retainer, AI coding agents, MCP server, Glydo, Dave Ebbelaar, Claude, Codex, ChatGPT

Full transcript (original language)

00:00 I’ve talked with more than 300 freelance tech professionals, and in this video, I’m going to share with you exactly what works, what clients are paying for right now, and how to get started. Because in the tech industry, there are two types of freelance models, and they are completely different, and most people only know either one of those. And now, if you’re new here, welcome. My name is Dave Ebelhar. I’m an AI engineer, and I’ve been freelancing in the tech industry for over 7 years already. And nowadays, I mostly just maintain the projects that I’ve built for my clients and build my own software. Okay, so the first thing that you need to know is the

00:31 difference between the two types of gigs, long-term contracts and project-based work. So, let’s cover what that means. So, first of all, long-term projects in the tech industry are quite similar to full-time jobs, like what you would see in the tech industry in general. So, this could be an AI engineering role, a software engineering role, a data science role, or any of the tech roles that are available, but then, this is the big difference, on a freelance basis. And you typically get those through either like a recruiter or an IT staffing agency. They’re mostly

01:02 full-time, 40 hours a week, sometimes 32 hours a week. And the pay for this as a freelancer is generally alt- always based on an hourly rate or a day rate. Now, on the other hand, we have project-based work. Now, this is what most people think about when you think of the idea of freelancing, right? Let’s say you get asked to build an AI assistant for real estate brokers. So, you decide we need 2 weeks to build this. Uh you’re agree on a 12K fixed price project for this, and then 50% is due upon signing, and 50% is due upon

01:33 delivery. That is typical project-based work in the tech industry. Now, why is it so important to understand that? Well, let’s take the following image over here. So, let’s paint a picture of what a year as a freelancer looks like. Let’s say you are a freelancer, and you want to work on long-term contracts. Here’s what your year could look like. You could go 1 year with just a single client that pays you hourly or on a day rate and can pay you for up to 40 hours a week. Now, if you do the calculations and the math on here, you can often earn

02:06 two to three times as much as regular employment for almost the same role. Now, you might be wondering, but yeah, how is this different from a normal job? Well, there are so many difference between that. So, first of all, like I said, the pay. Like the pay should already be a good enough motivation, but also you are often not really involved in any company politics, so you can just focus on the work. It doesn’t have to be 12 months. It can also be three months or six months and then you can go to another company, you can work on another project. So, there is so much more to

02:37 this and I haven’t even covered the whole concept of starting your own business, business expenses, tax deductions. You have so much more freedom and flexibility when you structure your career in tech like this. Now, the other type of work is project-based work. And if we compare a freelancer in tech who just does project-based work, it would look it generally looks something like this. So, you need to do a little bit of selling because you need to find a company to work with, right? On a project like end-to-end. And then you land one, you

03:09 do the work, but then it’s not always the case that the next project is already lined up because the increments are just like smaller here. It’s like two weeks, three weeks, maybe a month. So, you need to stack more projects back to back. So, that means more outreach, more selling, more sales calls, more proposals. Whereas one long-term contract can be acquired through a recruiter or staffing agency and of course that also doesn’t happen overnight, but if you land one, you’re good for like a long duration and that usually also gives you enough time ahead

03:41 to already start looking for the next one when your your exit there is coming up, right? So, this is so important to understand because most people currently have a job in tech and they think about starting as a freelancer and they think about project-based work only. They think about I’m going to start small. I’m going to stack project next to my full-time job and then as I build up momentum and as I build up a of enough clients, I’m going to take the leap and I’m going to go full-time. But most people never make it full-time if you

04:12 just do project-based work. That’s based on my own experience and that’s based on talking to over 300 freelancers in tech industry because here’s something that not a lot of people tell you. 80% of the revenue from freelancers in tech that do this full-time for a long time, 80% of that revenue comes from long-term contracts. Not the chatbot that they make for a local business, not the type of jobs and gigs that all of these people on YouTube tell you about like how to start your AI automation agency and how to service

04:44 clients. In the tech industry, people make money by establishing long-term contracts and long-term relationships with enterprises and working on that. So those are just the facts and that is super important to understand. So what if you are someone who decides like, “Hey, this sounds interesting. I want to work on those freelance long-term contracts. What type of profile do I need for that?” Well, it is pretty similar, almost similar to a normal role as well. So what are these companies looking for? Well, think about it. These

05:14 are enterprises and you are going to work inside an existing team. So you usually need to bring at least like three or five years of experience for the type of work that you want to work on. So if you’ve been in tech for a couple of years, you can apply for these types of contracts. So if you want to go AI engineering, for example, stuff that I did is you would need to prove that you have been working with LLMs for over three years, ideally like more traditional work before that before LLMs. So you need to show your track

05:45 track record. You need to show, for example, that you’re capable with fast API Python rag. So, this is very similar to a job application, right? And the conversation is usually around your CV. What do you bring to the table? What is your experience? And then they see if you’re a good fit for this. Now, what do enterprises hire contractors for in 2026 right now? And if you’re in software, the hottest thing right now to get into is shifting kind of like more towards AI engineering, where you integrate LLM features into existing products. This is

06:17 the most in demand right now. And when it comes to data roles, a lot of the focus right now is on building pipelines that feed AI. Any machine learning is still deploying, monitoring, evaluation, not much has changed there. And then of course, AI engineering, where you can think about rag, building multi-agent systems, and also evaluations. So, this is all very similar to the trends that we see in general in the tech industry, right? But now let’s switch to project-based work. And this is where it gets more interesting. So, first of all, beyond this whole concept of long-term

06:48 contracts versus project-based work, what you should also understand is the type of clients that you can actually work with as a solo freelancer. Because think about it. Do you think it’s realistic to do project-based work where you own an entire project and the outcome end to end? Do you think it’s realistic to do that on your own for an enterprise? Probably not, right? Because entire teams are already working on this. So, naturally, where you’ll then end up is working with small to medium-size enterprises. Anywhere from companies where it’s just a single

07:19 person, the founder, up to probably 200 employees or something like that. That is generally the range that we see where you as a solo freelancer can do meaningful project-based work. Now, what does that look like? And the answer to this is it’s actually quite simple. Whereas in enterprise, you’re part of highly complex teams building the latest and most advanced products and features. Whereas in small to medium-sized enterprises, it is the boring stuff, the stuff that every company has to deal

07:50 with and can now be automated to a great degree with AI. So, look at what we’re seeing over here. Invoices, emails, copy-paste between tools, writing quotes, weekly reports, the same questions over and over again, finding documents, data entry, planning, or just literally getting on top of the data from the company. This is the type of stuff that all small businesses without a tech team or developers on board struggle with. They simply have no idea how this world works and it’s never a

08:21 priority. So, this is the huge opportunity. This is where you with a tech background, and it doesn’t matter if you’re coming from data science, from ML ML, or from software engineering, you can work on these automations. Like it’s actually quite simple. It’s not rocket science. So, usually this involves around like what would we would categorize as back office automation. So, you take these invoices, emails, documents, phone calls, all of these kind of like artifacts, digital artifacts that flow through a company,

08:51 and you run that through automations and assistants. So, automations can be trigger-based, web hook based, running in the background, and the assistants are the chat applications, right? This could be directly into Claude or in Codex, or you could even build a custom chat layer. So, this is where the user is behind the wheel and is starting to ask questions. And the outcomes to that, so where does that go? Well, it’s all of the other parts of the business. So, a CRM, an ERP, a ticketing system, your accounting system, or directly back to the customers when it comes to support.

09:23 Back office automation. This is what small to medium enterprises need when it comes to AI automation. And the technology behind this, you can see like the middle smart automations assistant. Like that’s the like the technical layer. That’s which frameworks do you use, which language, which models? But the thing is the client doesn’t give a about that. They usually don’t even want to know. And if you talk about that, they’ll get they get all kind of like, oh, it’s like I I don’t understand this. So, it’s like so funny because if you go to the other world of like enterprise and long-term contracts, like

09:55 it’s exactly the other way around, right? It is specifically about which model you do we use, which language, like what are the architectural patterns and decisions that we’re making here. So, it is such a big difference and the thing is also like you can perfectly work on both at the same time. And it’s also what I’ve always done. But the like the mindset shift that you need to like learn and adapt when you’re coming from Big Tech and then start to enter into this world is night and day. And I often see in my communities, in my

10:26 freelance communities, the people that I work with, it’s often the people who are less technical that get results quicker. So, it’s often the people, for example, that come from research that struggle the most translating their work to the work that needs to be done for small to medium-sized enterprises. So, in the real world, how do you deliver these projects, right? Like I’ve said, it’s mostly just background jobs that run on its own on their own, like web hook based triggers. If this happens, then do that. You usually need a custom back end

10:57 for that. So, you you deploy it somewhere on the server. We just use Python back ends for that. We just put it on a VPS and then the other type of work is on demand, right? So, this is where the employee steps in, oh, I need to do this right now. And then they go to ChatGPT, they go to Codex, and they do the process. And this nowadays can either be a skill, just a skill, sometimes that’s enough, or it’s a skill plus an MCP server. And then setting up that custom MCP server, that’s usually where the magic is, right? Of course, there are a lot of

11:27 connectors out of the box already that you can use. But usually the best ones and the most effective ones are the custom MCP servers that you make and the skills around that that you help them engineer because that is still engineering and that requires an engineering mindset. So, here are some examples of what we recently worked on with our company. So, I’ll go over those quickly. A broker assistant, so this was for a real estate agent, so this is the chat agent category, right? A handbook chatbot for an internal team just to ask questions. A document checker, so this

11:59 was for any reports that were generated in the company, get them through an advance like spell checker and make sure the templates were on point. Creator analytics to track breakouts and to monitor essentially the competitors around them on social media. AI advisory, some startups and companies also just want advisory, not really anything built. Support automation, right? Connecting AI automation to the support and ticketing systems they’re already using, so an AI can create drafts and put replies there. Voice agents was for that same client we first did the email inbox and then we started

12:30 working on the voice agent layer from people that just call them. So, stuff that we recently built, project-based work, all small to medium-sized enterprises. So, now that you know the differences between the long-term contracts and the project-based work and have kind of like an idea of what those two different worlds are, let’s talk about what the best freelancers actually do. So, the freelancers that make the most money and they combine both. So, like I’ve said, 80% of the freelancers that do this for a long time and make a full-time living off of this, they at least have always have one type of

13:02 long-term contract going on. Usually, that contract alone, whether that’s one, two, or three a year or something like that depending on the length, will already generally put people at the six-figure mark. So, earning over 100,000 euros or dollars or more. Like with with just one contract sometimes or split up over multiple clients. But then, what they do, they stack smaller projects on top of that. And this is now easier than ever with AI coding agents because work is much more parallel, right? Before, before is like

13:35 3 years ago, if you were to do coding work, you would need to sit down, you would need to lock in for hours on end, weeks, days, months, and you would build a single project. Right now, it’s like you got one agent over here. Well, might as well kick off another agent over here, right? So, it gets easier for developers to take on more work at the same time. And the same is true if you’re watching this and you currently have a full-time job, you could even consider that blue long-term term contract as your full-time job,

14:07 right? It’s not freelancing work, but you can you can consider that. And then that’s naturally the best way to start, right? To do small project-based work on top of that. And then to put the cherry on top, maintenance retainers. And this is what we are mostly doing right now. And I keep saying we, by the way, because I have one business partner in the agency. I started as a solo freelancer, then turned that into an agency where at the highest point we were with around like six people. Then we scaled it down again to two because of AI coding agents. And

14:37 we can do more work right now with the two of us than what we could do last year with six people combined. So, that is really how those unit economics are also changing. But what we’re mainly focusing on right now is the maintenance retainer. So, we have been doing this for a long time already. So, naturally, there are a couple of clients who we’ve just done dozens of projects with. It starts with one built, and then you keep expanding on it, and we also host and maintain a lot of that software. So, with that, we now have maintenance fees.

15:08 And that puts us now in a really good spot where we could mostly live off just those maintenance retainers from the clients we’ve already worked with. And sometimes they get kind of like a new request, “Well, hey, can you build about new feature?” And we’ll scope it out. But it’s just like recurring revenue on a monthly basis that now allows us to build our own software. So, what I’m doing now with my time next to this YouTube channel and helping everyone in my community, like it’s mostly just like maintaining the clients that I’ve worked with over the past, and I’m like all in

15:38 on building my own software. So, now I can almost work full-time on this new software product that we’re building. It’s called Glydo. It’s a speech-to-text tool specifically for developers, and we are the fastest in the market. You can press a single button, talk to it, and the text will appear instantly. And I can guarantee you, if you’re a developer, you’re already using another tool, I highly recommend to check this out because I know you will like this better. If you want to try it out, there’s a link in the description, or I can give you 1 free month of Glydo on the pro version, so you can try it out completely for free, try all the new

16:09 features that we have, see how fast it is, and I can guarantee you you will cancel the other application that you’re using right now. So, this is what the top earning freelancers in the tech industry are doing. This is the secret. Long-term contracts, project-based work on top of that, plus maintenance retainers, and overtime work on their own software products to build IP and to build a real company that you can eventually sell and exit. Now then, one more thing, for people in my community and also people like you who watch my videos, I host free monthly sessions

16:42 where we dive into all things freelancing in the tech industry. So, these are master classes that every time go into a particular pillar of freelancing. So, think about client acquisition, how to find clients, proposals. And the next one is coming up in 2 weeks from the day of this recording, October 22nd, and this one goes all into freelancing in the age of coding agents. So, I’m going to talk about how you can most effectively use AI coding agents both for acquisition, for reaching out, but also for

17:13 delivering your actual projects in order to really build an army of AI agents that help you as a freelancer. So, if you really want to give this a shot, make sure to check out the link in the description and sign up for this free masterclass. It’s hosted on Zoom, it’s online. Like I said, it’s completely for free and you get to hang out with me and a whole bunch of other people. I will be there to ask your questions as well. It’s going to be fun. And if you are watching this video after that date, make sure to still check out that link because after the event, I will just redirect that same link to the next

17:45 upcoming event or free resources that I can share with you. All right, and that’s it for this video. If you found it helpful, please give a like and also consider subscribing. And then, I hope to see you on the event live with me to dive into your questions or the next event. And then, in the meantime, make sure to check out this video next. I will put one here which YouTube thinks you would like to watch next.