Harshit Tyagi

A few common industrial use cases of personalized ranking: Social Media: Facebook and Twitter use personalized ranking to organize the content in a user's feed, prioritizing posts based on their likely interest. Example: posts by friends are ranked higher. E-Commerce: Amazon and other online retailers employ personalized ranking to sort product listings, search results, and recommendations. Search Engines: Google personalizes search results by ranking them according to the individual's past search behavior, clicked links, and location data, aiming to provide the most useful and relevant results first. Online Advertising: Ad platforms use it to determine which ads to display, in what order, and with what frequency to a particular user.

Collaborative filtering enhances personalization across domains: 1. In e-commerce, it assembles outfits by analyzing user preferences for style coherence. 2. Streaming services utilize it for creating mood-based playlists, informed by users' emotional engagement with content. 3. Food platforms offer a culinary journey, recommending new cuisines favored by taste-alike users. 4. For e-learning websites, collaborative filtering can suggest a personalized pathway of courses that leads to the mastery of a subject, based on the paths taken by users with similar learning patterns and goals.

ML is pivotal in solving various business analytics problems by leveraging data to generate insights and facilitate decision-making: - ML algorithms like clustering can segment customers based on their behaviors and preferences. This segmentation allows for more targeted marketing, improving customer engagement and retention. - Predictive models can forecast sales, demand, and other key business metrics. For instance, using historical sales data, ML can predict future sales trends, enabling better stock management. - ML can optimize supply chain operations by forecasting demand, optimizing delivery routes, and managing inventory, leading to cost savings and improved efficiency.

Here are a few more ways of stand out from the crowd: 1. Data-Backed Portfolio - Use analytics and data visualization to showcase the impact of your projects. - Include key performance indicators (KPIs) for each project. 2. Solution-Centric / Impact driven Portfolio - Focus on the problems you have solved, emphasizing the real-world impact. - Include testimonials from clients / stakeholders. - Case studies 3. Open Source Contribution Highlight - Showcase contributions to open source projects. - Detail the issues you've tackled. - Include links to pull requests and the final, merged code. 4. Educational Portfolio - Include mini-tutorials or Colab that allow the viewer to replicate your results. - Offer downloadable resources.

I have a slightly different take on this, I recommend taking a contrarian approach by picking a niche project. It is harder to get to the top of a field with so much competition than positioning yourself in a way to create your own field / subfield. Why Niche Problems? - Less Competition - Unique Skill Showcase - High Impact - Easier to innovate since the problem itself is niche. The harder problem is "How to find niche projects". Here are a few ways: - Interdisciplinary Blend - Community Forums and Journals - Historical Data and older research papers or projects that had promise but were left unexplored due to technological limitations that may no longer exist. - Personal Experience in a domain. SMART goals are applicable here too!

There are a bunch of ways you can stay on top of the field, here are a few of the resources I follow in order of priority, preference and convenience: 1. Blogs and Newsletters: Meta AI, Google Research Blog, Netflix Tech Blog, The Sequence, The Rundown AI. 2. BooksJournals and Research Papers: arXiv, NeurIPS, ICML, ICLR Proceedings. 3. Books - Oreilly books are usually good and I am always on the lookout for new books which are more hands-on in nature. 4. Social Media: LinkedIn Groups, X(formerly Twitter) Lists, Reddit's ML Community. 5. Open Source Courses and Repos: fast.ai, madewithml blog by Goku, Coursera. 6. Podcasts and YouTube: Lex Fridman, Abhishek Thakur, Freecodecamp.