Episode 2: ML, Google Ads, Networking

Weekly Highlights

1. Machine learning course
2. Freelance Project Update
3. Events
4. People who inspired me this week

Machine Learning Course

I've completed Module 1 of the machine learning course and am now deep into Module 2, learning TensorFlow and completing hands-on coding labs. Module 1 introduced the Google Cloud Platform, teaching us how to use Compute Engine to build ML models, and also No code ML solution with AutoML providing practical exercises for running NLP models to analyze text sentiment.

Module 2 is when things really intensified. The course provides a comprehensive overview of TensorFlow, covering the Sequential API, Functional API, and diving into ReLU in great detail. To strengthen my understanding of gradient descent and loss functions, I watched several videos from StatQuest (absolute GOAT). As a side note, I used to watch StatQuest when I was doing my master's degree—the channel was relatively new then but helped me tremendously during exams. Revisiting these concepts now, I'm amazed to see how much his fan base has grown and how his videos remain the best resource for understanding ML concepts.

The TensorFlow labs were challenging, but I eventually got the hang of them and managed to complete them one by one. I have yet to go through the reading list, but I'm really happy with what I have learned so far.

Freelance Project Update

While desperately trying to uncover insights through machine learning, I jumped from k-means clustering to building a predictive model. I wanted to predict which low-performing campaigns would convert if I changed specific variables.

When adding ground truth values for my prediction model, I realized after multiple attempts that I shouldn't run any ML model yet. With my current data volume, simple descriptive models and tables easily identified clear winners and losers—I simply didn't have enough data to train a proper model. Instead, I decided to enrich the data through A/B testing. Though I've known about A/B testing since first learning about marketing campaigns, I've never actively implemented it. I understand the concept and use cases theoretically, but had no practical experience. I felt thrilled to reach a point where I intuitively recognized A/B testing as the logical next step.

I opened Google Ads to set up A/B testing but quickly realized I lack knowledge in creating Google Ads campaigns. Despite analyzing Google Ads data multiple times as a data analyst, I've never approached it from a marketing perspective. Setting up ads requires both art and science—exactly the intersection where I feel most at home. After watching some YouTube videos, I remembered I've been paying Coursera $50 monthly and could take a proper course instead. Learning Google Ads thoroughly would be valuable for hypothesis testing and model development. Unfortunately, this means pausing my ML workflow for 3-4 days while I learn Google Ads.

For those interested in the Google Ads Course, I've included the Coursera link below. Feel free to explore it.
Search Advertising with Google

Update on the course:

The course is actually quite light and fast-paced; I completed it in just 4 days. After finishing, I created campaigns with relevant keywords, organized each keyword into broad themes, and wrote about 15 headlines with 4 separate descriptions to test multiple display options for search campaigns. I also set up the followings:

1. Sitelinks
2. Callouts
3. Structured Snippets
4. Offers and promotions

I'm so glad I went through this course. Now I understand the inner workings of Google Ads, which has enabled me to view the data as a familiar entity rather than something external. It definitely enriched my knowledge of how campaigns in Google Ads work, why some ads succeed while others fail, and which KPIs are important to track.

Once my campaigns are properly set, I'm ready to start A/B testing. I was genuinely surprised by the sophistication of Google Ads; it already uses advanced machine learning models to do the heavy lifting of showing the right ads to the right users. Once the ad is live, the business owner's job is to fine-tune the website, creating a seamless user experience that makes customers feel valued and well-cared-for.

Room for thoughts and reflection

“There's a constant internal conflict I navigate between my passions and what I believe is the most pragmatic path forward. While I'm drawn to writing and exploring fields like YouTube and content creation, I feel compelled to dedicate my energy to machine learning because of its potential for future growth. This creates a sense of obligation that can sometimes feel at odds with my curiosity.

But I've had an important realization: my creative and technical pursuits aren't separate. My creative expression is actually fueled by my technical expertise. When I say "creative," I don't just mean making videos or writing blogs to showcase my work. I'm also talking about the creativity involved in problem-solving—the ability to look at a complex technical issue and find a unique solution. To find that solution, you have to become the technical expert. I now see that my creative and technical sides are two parts of the same whole.

I hope I'm not alone in thinking this. Do you ever feel the same way?”

Events

I've been submitting job applications daily, but as expected, they're disappearing into the application void or being rejected by automated tracking systems. With so many autofill features available, a single job listing can receive hundreds of applications, and resumes often look alike. It’s incredibly difficult and frustrating to stand out in this environment.

To combat this, I’ve started attending tech networking events, embracing the idea that connecting with people is key. My first event went well. Initially, I was unsure how to start conversations, but I mustered the courage to approach people and discovered it wasn't intimidating at all. Most attendees were simply there to discuss their work and have casual conversations.

At the event, I attended a fireside chat with Weaviate CEO Bob van Luijt, where I learned about open-source vector databases—an extremely valuable technology in the age of AI. I also participated in a panel on building AI agents, meeting a Cloud Architect from Volkswagen Group, professionals from SAP and Deepset, and a researcher doing cutting-edge work in computer vision.

The most relevant event for me was a workshop from Couchbase. I connected with so many talented data professionals who are excelling in their fields and feel I made some valuable connections there.

I've already signed up for a few more events and a hackathon, and I'm genuinely curious and excited about what might come from these opportunities. I'll share more in a future post.

People who inspired me this week

1. Inside OpenAI's Stargate Megafactory with Sam Altman - This video was eye-opening. It shows just how much major AI companies like OpenAI are scaling their infrastructure to meet future demand.
2. My 17 Minute AI Workflow To Stand Out At Work - taught me a great trick: how to write better prompts to get top-quality results from AI.
3. Brian Chesky, Co-Founder and CEO of Airbnb: Designing a 10-star Experience - His storytelling and the way he talked about designing a "10-star experience" for customers were just so cool.

See you in the next post!