DEV Community: Samia

Recipe to Transition from Dev to ML/Data/Applied Science Roles

Samia — Thu, 03 Feb 2022 17:23:13 +0000

As someone who has successfully transitioned from Software Engineering to Machine Learning Engineer and Applied Scientist roles, I get asked this question a lot so here is my recipe in under a min!
This has worked not only for me but also many other colleagues at Microsoft.

I hope you found this video helpful. You can help me reach out to others who might benefit from these insights by showing this post some ❤️ and subscribe to my YouTube channel for more content on AI & Data Science (and more).

P.S. Are there any topics that you would like me to cover next? Let me know in the comments on YT. Thank you and best wishes with your journey!🙂

Kaggle vs Real AI Product Development Life Cycle

Samia — Mon, 24 Jan 2022 20:39:46 +0000

Kaggle is great for keeping your data skills sharp, but if you haven’t been exposed to creating real world AI-powered applications, it can set some wrong expectations.

The AI/ML development life cycle consists of various stages and steps. It’s especially important to remember this and think holistically during your ML system design interview.

I hope you found this video insightful. You can help me reach out to others who might benefit from these insights by showing this post some ❤️ and subscribe to my YouTube channel for more content on AI & Data Science (and more).

P.S. Are there any topics that you would like me to cover next? Let me know in the comments on YT. Thank you and best wishes with your journey!🙂

AI & Data Career Pathways Deconstructed

Samia — Sat, 16 Oct 2021 00:04:08 +0000

Lack of clarity when deciding a career pathway, can be really stressful 😓 I have been receiving many questions and seeing a lot of confusion on this topic, so instead of continuing to answer individual questions, I decided to address them once and for all.

Developing an AI project is not a one person feat. It involves various technical roles and tasks. In this video I deconstruct them both and also go over skills needed and tools used in industry.

✔️Tasks covered: Data Engineering, Modeling, Deployment, Business Analysis, Infrastructure, Research
✔️Roles compared: Data Analyst vs Data Scientist vs Applied Scientist vs AI Researcher vs ML Engineer vs Software Engineer

With the hope that you find this concise yet comprehensive walk-through helpful and you are walking away with clarity and can make informed career decisions.

Best of luck with the journey ahead! 🌟

P.S. You can help me reach out to others who might benefit from it as well by showing this post some ❤️ and don't forget subscribe to my YouTube channel for more such content.

P.P.S. Let me know in the comments on YouTube, if there are any other topics that you would like me to cover next :)

5 Skills to Not Become Obsolete in the AI Era

Samia — Fri, 17 Sep 2021 13:50:34 +0000

In this post I want to talk about a slightly different topic than my usual, technical AI blogposts - something I believe is really important for us to succeed in the long term.

This article is based on my YouTube video on the same topic:

The world is changing fast. The Fourth Industrial Revolution is here. To stay relevant in the AI-era, we need to develop 21st century job skills, i.e., skills that cannot be replaced by robots and AI. This means embracing the change and then exploring and cultivating what makes us uniquely human.

1. Continuous Learning

Many people think that once they have completed a degree, i.e., they have done what it takes to get a job that pays them high enough, they are all set for life. Even if this was ever true, this is certainly not true in today's technology-driven world. The concept of upskilling is not new per se, but what has changed is that the pace of disruption is faster than ever - there is no one job for life anymore.
As the famous historian Yuval Noah Harari tells us in his awesome book "21 lessons for the 21st century":

Technological innovation and AI are going to accelerate at a pace we've yet to really comprehend so unless you are 80 years old or something, you will have to repeatedly reinvent yourself in the coming decades to survive that change - you'll probably change your job a number of times to not become obsolete.

This means that the value of continuous learning has become more important than ever before. So, unless we want to become obsolete in this fast changing world, both as individual and organizations, we need to keep up with it by learning to evolve rapidly.

Since the ability to learn rapidly is going to be such a huge game changer, you can go deeper on this topic here: ultimate system to learn anything smarter and faster

2. Critical Thinking

In the 19th century, when information used to be scarce, cramming students with lots of information made a lot of sense. Unfortunately, many schools still tend to place too much emphasis on doing the same. Today we have access to more knowledge on our fingertips than what was accessible to the President of the United State some decades ago - "traditional learning was built for a world that we don't live in anymore."

The essential skill needed in this century is to learn how to make sense of the vast amounts of information that bombards us on a daily basis. Our problem is not scarcity of information but all the misinformation that now exists. We need to learn how to distinguish between important information and irrelevant, between reality and fake news. We can do this by developing our critical thinking muscle.

To be clear, critical thinking has nothing to do with criticizing others. It is about making judgments based on well thought out and logical reasons. This means, when faced with a big decision, you put the brakes on, scan the horizon, develop well thought out and informed judgments and conclusions, and then only hit the accelerator. Basically, it's thinking about thinking.

Even if thinking critically doesn't always come naturally, it is a muscle that can be trained with time and practice. And training it pays off because it plays a huge role in how successful you are in life:

"It is in your moments of decision that your destiny is shaped." - Tony Robbins

3. Creativity

According to a LinkedIn study creativity was ranked as the number one soft skill. And even when analyzing both soft and hard skills, it still made it to the top skills employers search for.

We generally tend to associate creativity with things like composing music, and writing poetry, but creativity is much more than that. "Creativity is problem-solving with relevance and novelty", which explains why it tops the charts both among soft and hard skills.

We cannot compete with AI in recalling information, performing logical tasks to a tee or finding patterns from data, but when it comes to thinking outside the box, connecting dots, or designing experiences that play to the base elements of being human, AI does not meet the bar. A main reason for this is that we use our shared life experiences and emotions to bring creativity to life - a capability that machines lack and will continue to "struggle" with.

4. Emotional Intelligence

We will get to a stage where AI will get better at diagnosing diseases better than a doctor, and even recommend better treatments. But it cannot replace the value that a person sitting down with a patient and listening to them empathetically brings to the table. Feeling listened to by a doctor is not less important than their technical competence. And it will become an increasingly important skill for a doctor, unless they want to lose their job to a machine - there is great value to us in being seen, being heard, and relating to others.

We are far from a stage where AI can claim consciousness, emotions, empathy, or compassion. Understanding others, being able to motivate them and lead them with empathy are going to be the capabilities that will become more and more sought out over the next decade.

Begin to give importance to these skills the same way you focus on developing your technical skills. Develop emotional intelligence, become an outstanding motivator, leader and listener, so that when AI disrupts your job, you do not become obsolete.

5. Mindfulness

At the 5th Annual World Government Summit, over 4000 politicians, private sector leaders, policymakers, and experts, including the visionary Elon Musk, came together to discuss topics that will shape future governments. A radical vision of our future emerged which placed self-awareness and mindfulness at the center of human endeavor. Let's look at some of the reasons why:

Mindfulness is the core 21st century capacity, because it is about our biggest competitive advantage over machines: awareness itself. More often than not, we let our monkey mind run the show; we live in a default, distracted state. However, things we can do on autopilot, AI will learn to do them as well.

Mindfulness is the pillar supporting all the previous skills. Critical thinking, emotional intelligence or creativity, all these skills require us to understand our own subjective and collective experiences and to develop a sense of togetherness, empathy, and compassion.

Mindfulness also helps us develop the mental stamina to handle this unprecedented pace of change:

"Even if there is a new job, and even if you get support from the government to kind of retrain yourself, you need a lot of mental flexibility to manage these transitions. Teenagers or 20-somethings, they are quite good with change. But beyond a certain age - when you get to 40, 50 - change is stressful. And a weapon you will have [is] the psychological flexibility to go through this transition at age 30, and 40, and 50, and 60. The most important investment that people can make is not to learn a particular skill - "I'll learn how to code computers," or "I will learn Chinese," or something like that. No, the most important investment is really in building this more flexible mind or personality." - Yuval Noah Harari

Access to huge amount of data has given us power, but it didn't come with the wisdom to use this power sensibly. That requires conscious choices and awareness. While we are living mindlessly, without fully grasping this new world unfolding around us, technology is changing the DNA of our society. Our future generation will have to live with the consequences of our unconscious decisions today - let's not choose to allow technology to radically change our lives and society on autopilot; let's choose to use it to live more mindfully and create more fulfilling lives.

Summary

To succeed in the face of the inevitable disruption by AI, begin to focus on capabilities around continuous learning, creativity, problem solving, leadership, empathy and mindfulness.

If you found this article helpful, help me reach out to others who might benefit from it as well by showing it some love and subscribe to my YouTube channel for more smart content that can 10x your life 🔥

Happy Learning and Keep Evolving! :)

References and Further Reading

21 Lessons for the 21st Century

Ultimate System to Learn Anything Smarter and Faster

Samia — Thu, 02 Sep 2021 23:08:14 +0000

Learning efficiently is a game changer and it is going to become increasingly important in the coming decades because the world is evolving at an unprecedented pace.
As the famous historian Yuval Noah Harari tells us in his awesome book "21 lessons for the 21st century":

Technological innovation and AI are going to accelerate at a pace we've yet to really comprehend so unless you are 80 years old or something, you will have to repeatedly reinvent yourself in the coming decades to survive that change - you'll probably change your job a number of times to not become obsolete.

This means that if there is one skill we should master to create a great life, that is the ability to learn rapidly.
From learning technical skills, to getting my fitness game up, or learning skills just for fun, I am obsessed with learning and evolving myself. Today I want to share with you the system I use to learn anything the smart way.

This article is based on my YouTube video on the same topic - I highly recommend you watch it for a more detailed understanding. In this video I first describe a smart learning system and then I go through 7 techniques for effective learning. The idea is to cherry-pick the techniques you like the most and add them on top of the system to create your own personalized system for learning anything smarter and faster 🔥

For those of you who prefer reading, let's go through it, more briefly, in textual format.

The Smart Learning System

Step 1

Decide the skill you want and exactly define the level you want to reach - whether it's learning to play guitar or getting started with Deep Learning, what is the level of target performance you want to achieve? Be specific 🎯

"If you can't measure it, you can't improve it."

Step 2

The next step in hacking the learning curve is to deconstruct the skill into its basic, fundamental components and then using the 80/20 principle, a.k.a. Pareto principle - the goal here is to separate the 20% of learning material that will give us 80% of results 🧩

For example:

Say you want to learn to play guitar. There are some main chords (Em, C, D, G) that you can be used to play a lot of songs. So if you just learn those few chords and a couple of strumming patterns, you will be able to entertain people at the next party!
If you are getting started with Machine Learning then instead of trying to learn everything under the sun, you can focus on some core algorithms (like regression) and concepts that can be used to solve most of the problems.

(Pareto principle is fantastic, isn't it?!)

Step 3a

Practice with NO distractions from the outside world. From checking our emails every ten minutes, to scrolling through our Instagram feed or reacting to notifications, multitasking and not paying undivided attention to the task at hand prevents us from learning fast.
📌Focused practice. Undivided attention.

Step 3b

Practice with Consistency because there is no way to get around the fact that skills acquisition requires work!

If you can dedicate as little as 30 minutes per day for the next 30 days of consistent focused practice, that will be enough! Of course, to gain mastery you will need more hours, but 30 days will be sufficient to become good enough. Once you have achieved your target performance, you can repeat the 30 days loop to get to the next level.

"Atomic Habits" is a great book on learning to develop new habits and making them stick. For example, habit stacking tells us that we can couple the new habit with an old one to make it stick - you can practice the skill you are working on immediately after making your morning coffee or before brushing your teeth in the evening ☕

Step 4

Create fast feedback loops. You can't get better at something unless you know how to track your progress.

For example, if you are learning to code or preparing for an interview, you can test your progress on LeetCode. If you are learning to play an instrument, you can record yourself. If you are preparing for an exam, you can take some mocks to test the subskills. If possible, you can set up some checkpoints with a mentor or coach.

Step 5

Don't quit before or during the dip. There is dopamine hit when we begin something new. But once the honeymoon phase is over, we experience the "dip" and most of us quit. To not fall victim of the dip and give up too soon, pre-commit at least 30 days of daily practice to check how important that skill is for you 💪

"Anything worth doing is worth doing poorly at first." - Ray Congdon

7 Techniques for Smart Learning

1. Teach it to Someone Else / Feynman Technique 👩‍🏫

Feynman is a very famous physicist. His goal was to be able to explain even the most detailed and scientifically complex information to a child - if you can breakdown something intricate in a way that a child can understand, then you have a good grasp of the topic yourself.

2. Notes by Hand ✏️

Researchers at Princeton University and UCLA have found that when students take notes by hand, they listen more actively and they are able to identify important concepts much better. When we are taking notes on a laptop, we tend to be more mindless and distracted - ditch the computer or tablet in favor of the good old way of writing things down by hand.

3. Learn in Short Bursts of Time ⏳

Brief, frequent learning sessions are much better than longer, infrequent ones. Yes, those 90 minutes long lectures are not very effective! Our attention span has decreased too much. However, 5–10 mins of practice is not enough either. For the sweet spot, technique 4.

4. The Pomodoro Technique 🍅

This is a time management technique where you break down your tasks into 25 minutes of focused blocks of time - just the right amount of time to feel productive, without feeling the temptation of getting sidetracked.
You can find many apps and online timers to help you track your pomodoro sessions.

Aim for 2 pomodoros a day for 30 days for the skill you want to acquire and see the magic happen!

5. Power of Mental Spacing 🧠

Research tells us that we are able to remember concepts better if we learn them in multiple, spread-out sessions because when we repeat the information over a longer interval (e.g., a few days or a week later), it sends a stronger signal to the brain that it needs to retain that information.

6. The Sleep Sandwich 😴

Our brain needs time to recharge and file all of the information we fed it into proper "memory bins". Storing memories is one of the most important functions of a good night's sleep while cramming and all-nighters disrupt this active storage of content into long-term memory. So, prioritize downtime to learn more efficiently.

7. Environment and Dedicated Spaces 🌱

Immerse yourself in an environment most conducive for learning the skill you are working on.

For example, when learning French, if you can manage a trip to France, that will accelerate your learning. Or say you want to transition from Software Engineering to Machine Learning/Data Science roles. While you are still in the skills acquisition phase, and haven't reached the stage where you can switch roles, you can start by getting your foot in the door (under your current role) by being part of a team that focuses on Machine Learning to hack your learning curve.

Generally, creating a dedicated space for learning is also very effective - use your bed for sleeping, not studying, because our brain creates associations with our physical spaces over time.

In summary, the idea is focused consistent practice with 80/20 rule to mark the territory and then combining this learning system with a few learning techniques of your choice, like Pomodoro and Feynman, to create your personalized system for learning anything the smart way.

Let me know in the comments which skills you are going to learn :)

If you found this article helpful, help me reach out to others who might benefit from it as well by giving it a thumbs up and subscribe to my YouTube channel for more smart content that can 10x your life 🔥

Happy Learning!