The Entrepreneur Forum | Startups | Entrepreneurship | Starting a Business | Motivation | Success

Ideas using Machine Learning?

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
I have some knowledge in machine learning, image processing, and deep learning. I was hoping there was a way I could use my knowledge of ML to work for myself and scale it and generate passive income, as opposed to working for a company or as a freelancer

The only action I've really taken is starting a couple courses on Udemy. I was thinking I could make start a Youtube channel in which I go over machine learning projects that are simple and easy to understand for beginners

Or maybe I'm better off just starting my own web business (for example, an e-commerce site) and learning marketing/sales and hiring/outsourcing a machine learning freelancer?

Anyone else here with some knowledge of machine learning? What are some good ways to use it to get in the fastlane?
 
Last edited:

Become a Fastlane INSIDER to view the forum ad free.

Jeff Noel

Gold Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
I've Read UNSCRIPTED
Speedway Pass
Oct 26, 2018
524
1,043
369
Quebec, Canada
jfnoel.com
What are some good ways to use it to get in the fastlane?
The fastlane is a mindset. You can find good ways of using Machine Learning pretty much everywhere right now. Just listen, find real needs and provide value to people. Machine Learning is just starting... and it can be applied to pretty much any niche.

Don't chase money, provide value.
 

hydemx

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Jul 20, 2018
28
56
28
The fastlane is a mindset. You can find good ways of using Machine Learning pretty much everywhere right now. Just listen, find real needs and provide value to people. Machine Learning is just starting... and it can be applied to pretty much any niche.

Don't chase money, provide value.
This. Do not try to find a way for making money with Machine Learning. Find a way to solve a problem with Machine Learning, and money will follow.
 

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
Ok, so once I find a need and can provide value, is someone like me who already has years of experience with programming in machine learning (but nowhere near expert-level and there's tons of people who can program better than me. I also have no web programming skills) better off working on the machine learning/technical side? Or am I better off just outsourcing that work and devoting my time to sales, marketing, etc?

I saw some of the previous threads here about whether to outsource the coding or learn it, but those seem mostly for people who have no prior programming experience. I've already devoted years to programming
 
Last edited:

lowtek

Platinum Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
Summit Attendee
Speedway Pass
Oct 3, 2015
1,482
4,621
1,130
37
Phoenix, AZ
Subscribe to me on YT (link in signature) - yes a shameless plug.

You can freelance right away. Head on over to my channel and check out the machine learning freelancer series, and then check out @Lex DeVille content for some more in depth material on how to land gigs as a freelancer.

@Remiremi has also had success doing some freelance stuff in ML, and that has translated into other opportunities for him.

YT is a tough grind. I've been doing it off and on for almost two years and just crested 1,000 subs. I quit my last gig and now I'm doing YT full time as I build an audience to sell material to.

I'd recommend Fast.ai over Udemy courses, in general, but if you want to go the Udemy route, check out Lazy Programmer's stuff. It's quite good, though I don't care for his coding style.
 

lowtek

Platinum Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
Summit Attendee
Speedway Pass
Oct 3, 2015
1,482
4,621
1,130
37
Phoenix, AZ
I just read that you don't really want to freelance, after writing my above post. My bad.

Well, I'd say that you should freelance, or get a job, first, so that you can see what the market actually needs. Trying to guess and launch something isn't a great idea, unless you have some other product that people are paying for that you can improve with ML / DL.
 

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
I just read that you don't really want to freelance, after writing my above post. My bad.

Well, I'd say that you should freelance, or get a job, first, so that you can see what the market actually needs. Trying to guess and launch something isn't a great idea, unless you have some other product that people are paying for that you can improve with ML / DL.
I already have a job, but it's image processing (with a little deep learning) related. I also have freelanced a little bit in the past. How is freelancing part of the fastlane? I don't see how it generates passive income and scales
 
Last edited:

lowtek

Platinum Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
Summit Attendee
Speedway Pass
Oct 3, 2015
1,482
4,621
1,130
37
Phoenix, AZ
I already have a job, but it's image processing (with a little deep learning) related. How is freelancing part of the fastlane? I don't see how it generates passive income and scales
Generating additional income, building contacts, broadening your skillset, and engaging the marketplace has an infinitely higher probability of leading to a fastlane business than sitting around on your own trying to come up with ideas that generate passive income and scale.
 

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
Generating additional income, building contacts, broadening your skillset, and engaging the marketplace has an infinitely higher probability of leading to a fastlane business than sitting around on your own trying to come up with ideas that generate passive income and scale.
Ok that makes sense. I did freelance a bit in the past. I had a few different clients I found from Upwork and other sites, but those were very short-term projects
 

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
Oh, I forgot to mention that I can't even use Upwork anymore. Even though I completed projects with a couple different clients there, I was banned from the site without an explanation. I contacted them numerous times and they declined to give me an explanation for the ban
 

Become a Fastlane INSIDER to view the forum ad free.

Remiremi

Bronze Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
I've Read UNSCRIPTED
Feb 18, 2018
71
113
130
25
Paris
Hey @Rodmunch1
Although freelancing is not fastlane, as @lowtek said freelancing is a good way to get the grind going, also you can generate much more cash than a 9-5 and be in control of your hours meaning you can make room for rest or for any other side projects you like. Plus some other collateral benefits like networking, skills etc... I am not gonna extend on that as @Fox already put the concept into words here in his disclaimer. : How to Learn Code, Start a Web Company, $15k+ per month within 9 months

About Ideas I will suggest teaching ML locally, there's a huge demand for ML professional face to face formations and very few teachers available, for example, now that I made the jump and am teaching ML three days a week, I am approached by other opportunities that I have to turn done because they don't quite fit into my schedule. (You can't be two places at the same time)

A day of formation can be worth 500 to 3000€ to you. If you get good at it before the bubble explode, you can definely stack up some cash and contacts, and then move on with higher success rate to a fastlane idea.
 

Rodmunch1

New Contributor
Read Millionaire Fastlane
Apr 5, 2019
8
1
4
Hey @Rodmunch1
About Ideas I will suggest teaching ML locally, there's a huge demand for ML professional face to face formations and very few teachers available, for example, now that I made the jump and am teaching ML three days a week, I am approached by other opportunities that I have to turn done because they don't quite fit into my schedule. (You can't be two places at the same time)

A day of formation can be worth 500 to 3000€ to you. If you get good at it before the bubble explode, you can definely stack up some cash and contacts, and then move on with higher success rate to a fastlane idea.
How do you find these face-to-face teaching opportunities?
 

Remiremi

Bronze Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
I've Read UNSCRIPTED
Feb 18, 2018
71
113
130
25
Paris
Really depends of your geography to be honest.

In Paris, ML is fashionable so I guess those opportunities are easier to come by, what worked for me to get the foot in the door was a mixture of two things
A) Having a profile on a local freelancing platform
B) Volunteering to teach AI for free and organizing AI meet-ups.

(A) gave me the opportunity and (B) the credentials to seize it.
 
Last edited:

Scalr

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Apr 3, 2019
23
31
26
Palo Alto, CA
Anyone else here with some knowledge of machine learning? What are some good ways to use it to get in the fastlane?
I am working in a company that uses ML and I am thinking along the same lines.
Check Upwork Jobs to see what companies are looking for.
Some want you to set up the whole stack for them (AWS, Tensorflow etc) and show that you can run some models. I think this could be something companies will want more and more in the coming years. Some will want you to train models. Or are looking for datasets (labeled).

I also think some companies first need to see what can be done using ML. Many have no idea and need to see the light first. Like how you could train models for malls (maybe not a good idea with malls dying, idk) but use their video footage & computer vision to predict customer behavior, how changes to the mall change customer behavior.
Maybe: How can you increase customer time spent in malls by adding some decoration?

Go to mall/store owners find out what they would like to know.

It is probably also not the time yet to approach any company. Some of them just realized they need to organize their data better and are probably investing there first.

But I agree, there is a lot of money to be made and it’s time to really think about it. Thanks for starting this thread. :)
 

Scalr

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Apr 3, 2019
23
31
26
Palo Alto, CA
Here is another one:

Car repair shops probably have a dataset about customers - miles driven, repairs done etc. Using machine learning over several datasets you could predict upcoming issues for their customers vehicles. Using VIN and miles, fuse this with a database that has known issues per vehicle after x miles of driving. Surface to the shop owner.
Prepare an email based on all this information, offering personalized information and discounts.


+ based on past customer visits, run a model that estimates how many miles the customer drove until now and should get this pretty accurately.
 

404profound

Platinum Contributor
I've Read UNSCRIPTED
Speedway Pass
Aug 27, 2017
1,271
2,598
658
Desert of Desertion
I am working in a company that uses ML and I am thinking along the same lines.
Check Upwork Jobs to see what companies are looking for.
Some want you to set up the whole stack for them (AWS, Tensorflow etc) and show that you can run some models. I think this could be something companies will want more and more in the coming years. Some will want you to train models. Or are looking for datasets (labeled).

I also think some companies first need to see what can be done using ML. Many have no idea and need to see the light first. Like how you could train models for malls (maybe not a good idea with malls dying, idk) but use their video footage & computer vision to predict customer behavior, how changes to the mall change customer behavior.
Maybe: How can you increase customer time spent in malls by adding some decoration?

Go to mall/store owners find out what they would like to know.

It is probably also not the time yet to approach any company. Some of them just realized they need to organize their data better and are probably investing there first.

But I agree, there is a lot of money to be made and it’s time to really think about it. Thanks for starting this thread. :)
Also of note, I see tensorflow now has a JS implementation, so web-based ML is now feasible in the browser (I can hear every ML engineer screaming "don't do it!"), and I mostly agree - JS being single-threaded means slowww processing for large datasets. However, there are ways around it, like asynchronous loading and chunking. Just mentioning it in case anyone has JS knowledge already and wants to try out tensorflow.
 

lowtek

Platinum Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
Summit Attendee
Speedway Pass
Oct 3, 2015
1,482
4,621
1,130
37
Phoenix, AZ
Here is another one:

Car repair shops probably have a dataset about customers - miles driven, repairs done etc. Using machine learning over several datasets you could predict upcoming issues for their customers vehicles. Using VIN and miles, fuse this with a database that has known issues per vehicle after x miles of driving. Surface to the shop owner.
Prepare an email based on all this information, offering personalized information and discounts.


+ based on past customer visits, run a model that estimates how many miles the customer drove until now and should get this pretty accurately.
It's an interesting idea, but it depends on too many unknowns.

For instance, if a person does their own oil changes, it may appear that they are about to grenade their engine if they don't visit a dealership for a change, but they're probably fine.
 

Scalr

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Apr 3, 2019
23
31
26
Palo Alto, CA
It's an interesting idea, but it depends on too many unknowns.

For instance, if a person does their own oil changes, it may appear that they are about to grenade their engine if they don't visit a dealership for a change, but they're probably fine.
Absolutely, the edge cases. I think it depends on how you position your email. So it is not too "creepy" but tells customer if they need it, you are there for them.

Like what Target did with pregnant moms, they were able to predict that on buying behavior and offered huge discounts to them for car seats, diapers etc. This way they wanted to convert them to become regular customers. As childbirth is such a life-changing event, might as well change shopping habits.
 

Scalr

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Apr 3, 2019
23
31
26
Palo Alto, CA
Also of note, I see tensorflow now has a JS implementation, so web-based ML is now feasible in the browser (I can hear every ML engineer screaming "don't do it!"), and I mostly agree - JS being single-threaded means slowww processing for large datasets. However, there are ways around it, like asynchronous loading and chunking. Just mentioning it in case anyone has JS knowledge already and wants to try out tensorflow.
This might actually be super interesting.

There are also lots of Automation Startups raising tons of money recently. What they offer is basically a bot that watches user behavior and then finds tasks that can be automated based on that. (financial, accounting, support, internal communications etc) - pure machine learning. "Automation Anywhere" is one example. They raised $300m from Softbank. And corporations are just realizing what all can be automated.

Smaller agencies doing this for them might have a shot too, maybe? No $300m needed to start something like that. But some knowledge in ML.
 

404profound

Platinum Contributor
I've Read UNSCRIPTED
Speedway Pass
Aug 27, 2017
1,271
2,598
658
Desert of Desertion
This might actually be super interesting.

There are also lots of Automation Startups raising tons of money recently. What they offer is basically a bot that watches user behavior and then finds tasks that can be automated based on that. (financial, accounting, support, internal communications etc) - pure machine learning. "Automation Anywhere" is one example. They raised $300m from Softbank. And corporations are just realizing what all can be automated.

Smaller agencies doing this for them might have a shot too, maybe? No $300m needed to start something like that. But some knowledge in ML.
Really if you know a few classification methods and how to write an async function in node you could roll it in tensorflow using their documentation. I can think of use cases that would even eliminate my current position lol.
 

Become a Fastlane INSIDER to view the forum ad free.

NovaAria

Contributor
Read Millionaire Fastlane
Jul 18, 2018
36
71
112
Just dropping by to say that, as someone who's from a completely different field and with no knowledge on programming and ML, I am absolutely loving this discussion. I can smell the gold brewing here.
If my understanding of the field is correct, the biggest problem that major companies are facing when they try to embrace AI and ML is that they collect loads of "dirty" data that needs labeling, sorting, and preparing before it can be thrown into models?
Is this something that their in-house data scientist/analyst or ML agency does for them? Sounds like a waste of time and money if your engineer is busy filling excel sheets on your dime, and... If there is waste, there is a problem that needs solving...
Is my thinking correct?
 

404profound

Platinum Contributor
I've Read UNSCRIPTED
Speedway Pass
Aug 27, 2017
1,271
2,598
658
Desert of Desertion
Just dropping by to say that, as someone who's from a completely different field and with no knowledge on programming and ML, I am absolutely loving this discussion. I can smell the gold brewing here.
If my understanding of the field is correct, the biggest problem that major companies are facing when they try to embrace AI and ML is that they collect loads of "dirty" data that needs labeling, sorting, and preparing before it can be thrown into models?
Is this something that their in-house data scientist/analyst or ML agency does for them? Sounds like a waste of time and money if your engineer is busy filling excel sheets on your dime, and... If there is waste, there is a problem that needs solving...
Is my thinking correct?
Yes, and the other part of it is building models that "learn" on their own. That means with the cleaned data the ML engineer is "training" an algorithm to predict future outcomes based on statistical analysis of the data it is fed. In finance a model can be trained to predict stock performance based on large amounts of passed data being analyzed and used to train the machine. In healthcare a machine can be trained to find tumors based on the images it had been fed previously, and so on.
 

Scalr

Contributor
Read Millionaire Fastlane
I've Read UNSCRIPTED
Apr 3, 2019
23
31
26
Palo Alto, CA
To keep the inspiration going:
In transportation a model gets trained to detect what happens inside & outside the vehicle. Fusing it with a whole bunch of other sensors (GPS, Gyroscope etc.) Yes, this is used for self-driving cars.

In finance it is also used (by banks) to analyze over millions of transactions to detect fraud. Like when Chase messages you that there a suspicious activity, it is because there an anomaly to your usual purchase behavior.

Some more thoughts on where else could it be used:
Dating, analyze over big amounts of data to find the right person online. BUT this person doesn't even need to sign up for your service because all kinds of data is publicly available. :)
Based on location, hangout spots, interests or the type of filter they use on their instagram photos.
Creepy? Probably.

Here is another good one:
We know that all companies collect stuff about you to analyze. They will not always share this data with you. How about a user based ML - tell me what other companies know about me based on tracking my behavior on MY OWN machine.

Maybe even, based on that we compare and analyze your chances of success in certain areas of your life based on data from other users. (anonymized) - no need to fill out a 10 question survey. Just install our tracker and come back after a week.

We tell you everything you didn't know you are. :)
(Yes, I am reading the Boron letters right now and am just getting started on that.)
 

The-J

Legendary Contributor
EPIC CONTRIBUTOR
Read Millionaire Fastlane
I've Read UNSCRIPTED
Summit Attendee
Speedway Pass
Aug 28, 2011
3,467
7,771
1,736
Ontario
Machine learning is dope. For you, I'd recommend finding an interesting project that people further along than you have already started and joining on.

I don't know anything about ML/DL except the very basics of how they work but I've already got a few ideas. I can't execute on them right now, though.
  • ML ad creative optimization. Take data from Facebook or other ad network sources and use them to make ad creatives. Still images and text only, video is infinitely more complex. Create this technology, patent it, and sell to a large company.
  • ML financial modelling SaaS for small companies. Have a program that creates robust financial forecasting models for businesses and sell as a SaaS.
  • Improvements to existing NLP algorithms or programs to create chatbots that actually F*cking work. (I tried to create a chatbot for an 8 figure company and it went over like a lead balloon)
ML is not just automation. It's something deeper than that.

Man, ML is so cool. I need to devote more of my time to learning about it.
 

NovaAria

Contributor
Read Millionaire Fastlane
Jul 18, 2018
36
71
112
I agree.
Recently, something interesting happened that people who follow the gaming industry have probably heard of. A game called Anthem, published by a certain company called EA, was launched. Users eventually discovered that the game had certain facets of ML integrated (rather poorly, this was probably a first attempt). EA was trying to maximize playtime by manipulating drops, rewards, damage and game difficulty by tracking user behavior and adjusting the game on the fly. This went poorly as the game felt extremely weird and simply unfun.
Interestingly, though, once people discovered this integration they started gaming the system. For example repeatedly closing and reopening the game to trigger rewards, or taking long breaks to let the game encourage a play session by enhancing the drops.

The human brain wins again... This time.
 

BelgianDude

New Contributor
FASTLANE INSIDER
Read Millionaire Fastlane
I've Read UNSCRIPTED
Mar 6, 2019
8
8
19
A few problems that can be solved (or have already been solved?) with ML:
- Amazon recommends relevant items during or after a purchase. It is highly likely that ML is also used here. One of my previous employers also wanted to recommend relevant articles during or after a purchase. However, there was not a single employee with ML knowledge. It was decided to introduce the feature and someone had to micro-manage all products (link product a with product x and y, link product b with product z, ...).
- In many warehouses a lot of time is lost during picking due to non-optimized product locations. There are already solutions for this where once again an employee has to micro-manage everything. (For example, check which products are sold the most each day/week/month and group them at the beginning of the warehouse).

To find such problems, place yourself in the shoes of an employee in a random niche and think about the daily tasks. If the employee needs to micro-manage something, this can often be automated using ML.
 

FierceRacoon

Contributor
Jun 1, 2019
40
42
97
selfimprovementmindset.com
I've had a successful startup exit thanks to a recommendation system I had developed. I do not recommend this subfield of machine learning, as the value you are generating is limited. It is great if you work for Amazon, Netflix, and Spotify, but very low for just about everybody else, including Facebook, Google, Linkedin, and Microsoft. It is not that the value of recommendations is low, the value of better recommendations is low. (You may think that the recommendations provide a conversion lift, but very often they convert people who would've converted anyway. Imagine a recommendation system recommending posts on this forum. Useful? Yes. Critical? No. If you are building system, by the Law of Effection, your systems have to be where the money is. Your ML has to be mission-critical for the business to make money, if you want to have upside and to be treated well. )
Besides, the skills are poorly transferable. It is less rigorous than image processing, less lucrative than creating trading strategies; you don't learn as much about marketing as you would from directly optimizing funnels and A/B testing campaigns. It's just a subfield offering limited opportunities, in my view.

Besides, Netflix admitted at a conference that the metrics used "do not allow to distinguish between an enhanced quality of life and an addiction." I've asked a bunch of people at a RecSys conference, and everyone building recommendation systems admitted that they are basically trying to get people addicted to the Internet, wasting their lives online. Intentionally. How's that for a sense of purpose?


That aside, if you are good, you can join an early-stage ML startup and get an upside in the form of stock/ equity. Or you can start one. If you are not an expert, become one --- or forget it. If you want to go that way, PM me, send me your resume, and I might have some advice.

And there are sooooo many biggest problems when it comes to the data, I can't even begin. There's collecting this data. There's accessing it. There's finding competent tech people who can do something with it. Then there's finding competent business people who can decide what to do with that, because tech people don't care about $. Then... don't get me started.

In terms of predicting stock performance, it's a bit harder than that. Here is what it's like: you've had a trading model that used to work, then it suddenly doesn't work and starts losing money. You spend 10-12 hour days at work, trying to figure out how to fix it, before you lose your bonus if your strategy does not start performing again asap, and then your job, and the efforts of several last years with that. You need a solid tech background, AND THEN you need to commit another 5 years of overtime work if you want to get real results creating trading strategies that make money.
I've rejected it for myself essentially because it's a SlowLane, and if you fail, you get crappy skills out of it. If you fail to get rich as a sales person, at least you get certain social skills. If you fail at creating trading strategies, your last 5 years of ultra-specialization are of little use, unless you want to spend the rest of your career doing something similar. And some specialize even more, by fine-tuning the strategies to custom hardware. Even owning a trading firm is NOT a Fastlane, it's a lifestyle.
 
Last edited:
  • Like
Reactions: Kid

Post New Topic

Please SEARCH before posting.
Please select the BEST category.

Post new topic

Fastlane Insiders

View the forum AD FREE.
Private, unindexed content
Detailed process/execution threads
Monthly conference calls with doers
Ideas needing execution, more!

Join Fastlane Insiders.

Top Bottom