What (exactly) is Mindrift?

Mindrift is a platform that connects our clients and experts. Many of our clients are tech companies or are interested in creating their own GenAI models. 

An invitation to the Mindrift platform isn’t a job offer, and we don’t expect you to put in a certain amount of hours or rely on it as your only source of income. To put it simply, it’s a side gig—albeit one that requires a certain level of expertise and domain knowledge. 

What the customer wants

Like most opportunities within the gig economy, the steadiness of work depends on the clients. 

Let’s say you’re working as a driver for any one of the large ride-sharing companies. On a sunny Friday afternoon, you might only make a few dollars because the demand is low. But on a rainy Monday evening, during rush hour, you might be overwhelmed with calls because the demand would be very high. 

Your experience on Mindrift works in a very similar way—sometimes our clients require a large number of tutors for a very broad project. Other times, they might request a select group for projects with a narrow scope.

Recently, we ran a survey for our Discord community of freelancers to better understand their thoughts and feelings in regards to downtime. A large majority (88%) have experienced downtime at some point in their collaboration with Mindrift, but it’s not always a negative event. 

Results from our Project Downtime Experience Survey.

“I view the pauses as gifts of extra time,” says Jo, an Editor at Mindrift. “Those lulls in between projects have been used to spend time with family, gain new skills, or rest. Although it can be viewed as frustrating, it adds to the flexibility of freelance work.” 

Hyper-speed evolution

Another factor to consider is how fast the AI industry moves. 

About a year ago, it seemed like almost every AI tutoring platform was looking for AI writing and editing freelancers. That’s because a lot of clients investing in AI were training their models on very broad domains. They mostly required prompt creation, output analysis and scoring, and output editing. 

While these positions still exist, AI models are becoming more sophisticated. Instead of general conversational prompts and outputs, clients are often focusing on specific domains like mathematics, accounting, or engineering. These new projects are often done on a much smaller scale with teams of highly specialized experts, reducing the availability of projects.  

We understand that the nature of the industry sometimes creates frustration or disappointment, especially during downtime or idle time. But a larger percentage of freelancers shared that downtime allowed them to pursue other interests, upskilling activities, and money-making opportunities.

But a larger percentage of freelancers shared that downtime allowed them to pursue other interests, upskilling activities, and money-making opportunities. 

“As a busy parent, I welcome these pauses between projects,” explained Marjorie, a former AI Tutor - Writer and now a QA at Mindrift. “I can invest more time in my home, my children, my hobbies, and other projects I contribute to. For example, I give lectures to undergraduate students at my local university on the importance of ethical and unbiased AI. Pauses in my Mindrift work allow me more time to plan engaging lessons and make new connections that then help me be a better data trainer for the next project that comes along!”

The project lifecycle: never simple, always unique

The best way to understand downtime or lack of projects is to shed light on the project lifecycle. Although there’s not always a standard process, potential clients approach us with specific needs, spanning industries like finance, healthcare, technology, and others. 

We asked Catherine Fedorenko from our Business Development Team and Petr Grigorev from the Product and Delivery Team to break it all down in detail.

Mindrift Team: Can you explain how Mindrift’s partnerships with clients work?

Petr Grigorev: Mindrift has a portfolio of actively engaged clients, but we obviously aim to grow our number of clients and expand our business. In practice, this expansion usually means we start with smaller pilots and projects, then have a pause for the client to evaluate our results. Once approved, we move into a full scale production stage. 

Mindrift Team: What does the preparation for a full scale production project look like?

Catherine Fedorenko: Each project is carefully scoped to align with the client’s goals and we incorporate these specifications into our final solution. This includes pipeline design, AI Tutor requirements, project taxonomy, and guidelines. The initial scoping process typically takes about two weeks, where we define project parameters and assess feasibility.

Mindrift Team: How do pilot projects differ from full scale production in terms of time?

Catherine: Pilots generally run for one to two months and are focused on proving concepts or validating a dataset’s effectiveness. Full-scale projects, however, may span three to six months, with larger dataset volumes ranging from tens of thousands to hundreds of thousands of examples.

Petr: It’s also important to mention that the modern AI world is changing quite quickly, thus project timelines are becoming shorter as AI developers are actively competing with each other. 

Mindrift Team: Can you go through the process of moving from a pilot project to a full production?

Petr: Once we receive confirmation from our client, our dedicated delivery team starts to build the pipeline and gather the resources we have. First, we try to involve experts from our existing talent pool. Second, we look for more experts, if the project requires it. Every project has an onboarding and production stage where experts gain an initial understanding of the project. 

Catherine: In terms of timing, recruitment and onboarding typically take one to two weeks, including knowledge tests and training. 

Mindrift Team: Can you give our freelancers any advice on boosting their chances of progressing to the full production project stage?

Petr: It’s extremely important to invest enough time in the initial onboarding and trial tasks. They should also pay attention to the feedback given during the project to maintain their success.

Mindrift Team: Lastly, can you give us any insight into pauses, or downtime, between projects?

Catherine: Because tasks are project-based, availability fluctuates depending on client demand and timelines. For instance, pilot projects may have a sudden influx of tasks for a few weeks, followed by quieter periods as clients evaluate results. Full-scale projects typically have steadier workloads but are still subject to pauses during client reviews or dataset iterations.

Why do we “hire” so much?

In short, we don’t. Here’s what we do instead.

Since our job is to connect our clients with the right experts, we require a large talent pool of individuals who might be suitable for the different types of projects our clients have in mind. This ensures that if, for example, Company A reaches out asking for 50 cybersecurity experts, we’ve already found, vetted, and prepared the right candidates for them. 

But what if Company A requests a cybersecurity project with an undefined launch date? We still have to fill our talent pool with experts to ensure they’re ready once the details are ironed out. This is why our experts and tutors sometimes experience idle time between joining the platform and getting an invitation to a project. 

This is also the reason behind some of our terminology. We try our best to be transparent about the entire process, using terms like opportunity rather than job in our ads. The Mindrift platform is an excellent way to:

• Gain new experience for your resume.

• Contribute to the advancement of AI. 

• Network and meet other professionals with similar interests.

• Work on paid projects. 

But it’s still a freelance opportunity, meaning you’re not going to get a stable work schedule, a salary, or employment benefits. 

“For me there have been times where I've been part of multiple projects at once and could work 50 or 60 hour weeks, and times where I've not been on any projects and have had to just wait and see. I think it's important to remember that it's both a freelance role and a platform where, by nature, the workload can vary greatly,” says Kamil, AI Tutor - Editor at Mindrift. 

FAQ: The TL;DR of it all

Is this a job?

No, an invitation to the Mindrift platform isn’t a job offer. Instead, we provide opportunities for experts to partner with leading clients around the world on projects that advance AI. 

Can this be my only source of income?

Most likely not. As with all freelance opportunities, the market ebbs and flows. Sometimes you’ll have the opportunity to consistently contribute on long-term projects, other times you might have periods of downtime. We wouldn’t recommend relying on Mindrift as your sole source of income.

Will there be downtime?

Most likely, yes. Whether it’s a few days, weeks, or months, downtime is a normal part of freelancing on the Mindrift platform as projects wind down and launch. Your skills and expertise might be a perfect fit for some projects, or the client might want to go in a different direction for others. 

How long does downtime last?

It’s hard to predict this indefinitely. A large majority of our freelancers (57%) shared that downtime sometimes lasts a few months. 

Am I guaranteed tasks/projects? 

We can’t guarantee tasks or projects. Even if you’ve got the best skills and expertise, it’s all a matter of client needs.