Day 22: Graph Traversal Concepts and Practice

🚀 𝗗𝗮𝘆 𝟮𝟯/𝟯𝟬 — 𝗗𝗦𝗔 𝗖𝗵𝗮𝗹𝗹𝗲𝗻𝗴𝗲 Day 23 and continuing deeper into graph problems. The interesting part about graphs is how many real-world systems they represent — networks, dependencies, routes, and connections. Problems start to feel more like exploring a map than just working through a list. Today’s focus was on using traversal techniques to explore nodes and understand relationships between them. 🔎 𝗗𝗮𝘆 𝟮𝟯 𝗙𝗼𝗰𝘂𝘀 • Strengthening DFS and BFS traversal • Understanding how to track visited nodes • Solved: ✅ Clone Graph ✅ Course Schedule ✅ Rotting Oranges Graph problems definitely require careful thinking, but it’s satisfying when the traversal logic finally clicks. On to Day 24 #DSA #Python #LeetCode #Consistency #SoftwareEngineering #ProblemSolving

Day 32 of #GeekStreak60: Simulating Radial Spread in a Binary Tree! 🔥🌳 Tackled the Hard-rated "Burning Tree" problem on @GeeksforGeeks today. Key Learning: Binary trees are strictly top-down structures. But when you need to simulate something spreading outward in all directions (like fire, or a network infection), you have to break that hierarchy. By mapping child nodes to their parents using a Hash Map, I effectively transformed the tree into an undirected graph. From there, I deployed a Breadth-First Search (BFS) to process the radial spread level-by-level, calculating the exact time required to traverse the entire structure in O(n) time. From studying high-level business frameworks to writing low-level algorithmic optimizations, the learning momentum is locked in! 🚀 #geekstreak60 #npci #coding #Algorithms #Python #DataStructures #GraphTheory #SoftwareDevelopment

🚀 Day 11 – DSA Daily Series Today’s Problem: Find Minimum in Rotated Sorted Array (LeetCode 153) Today I solved an interesting problem that involves finding the minimum element in a rotated sorted array. 🧠 Problem You are given a sorted array that has been rotated several times. The task is to find the minimum element in the array. Example: Input: nums = [3,4,5,1,2] Output: 1 💡 Approach I solved this problem using Binary Search. Key idea: • In a rotated sorted array, the minimum element lies at the rotation point • Compare the middle element with the rightmost element • If nums[mid] > nums[high], the minimum lies in the right half • Otherwise, it lies in the left half including mid By narrowing the search space, we efficiently locate the minimum element. ⏱ Complexity Time Complexity: O(log n) Space Complexity: O(1) 🔎 Key Learning Binary Search can be extended beyond simple searching and used to identify structural properties of arrays like rotation points. Continuing the DSA Daily Series — solving and learning one problem at a time. 🚀 #DSA #LeetCode #Python #Algorithms #BinarySearch #CodingJourney #ProblemSolving

🚀 Day 22 of Consistency | #75DaysLeetCodeChallenge 🧠 Today’s Problem : Daily Temperatures (#739) 💡 Key Learning: This problem introduces the concept of a monotonic stack, useful for solving “next greater element” type problems efficiently. ⚡ Approach: Use a stack to store indices of temperatures Traverse the array → While current temp > stack top temp → pop & calculate days difference Store result for popped indices Push current index to stack 🧠 Why this works: Monotonic stack keeps decreasing order Helps find next greater element in O(n) Avoids brute force O(n²) 🔥 Result : ✔️ Runtime: 116 ms (Beats 28.07%) 📈 A very important pattern for stack-based optimization problems. A big thanks to Shivam Singh, Nikhil Yadav & Akshat Tiwari for this amazing challenge 🙌 Consistency is compounding. Keep going. 💪 #Day22 #LeetCode #DSA #CodingJourney #100DaysOfCode #Python #Stack #Consistency

Building an LLM-based extraction service taught me one important lesson: Hardcoding prompts doesn’t scale. Moving from prototype to production, I realized the system around the model matters as much as the model itself. What actually helped make it usable across different use cases: • Decoupling prompts from code using a MongoDB-backed configuration layer • Supporting project-specific prompts instead of one fixed prompt design • Designing the service to handle varying extraction requirements without code changes • Using models like Mistral-7B for extracting structured outputs (JSON/fields) from unstructured documents The real challenge wasn’t prompt design — it was making prompt iteration safe, flexible, and maintainable in a production system. #GenAI #LLMs #SystemDesign #Python #AIEngineering

A handwritten digit classifier built from scratch using NumPy. The project focuses on understanding the core mechanics of a neural network by implementing the full pipeline directly, including forward propagation, ReLU, softmax, backpropagation, and gradient descent. It also includes training reports, experiments, inference, Docker, CI, and a small demo to make the workflow easier to run and inspect. Latest MNIST run reached 95.78% test accuracy. GitHub: https://lnkd.in/gMS7a99g #machinelearning #numpy #neuralnetwork #python #mlengineering

In the past years, foundation models have been extensively utilized in time series forecasting, with models like TimeGPT and TimesFM gaining significant attention. Kairos is a flexible and efficient foundation model designed to handle the dynamic and heterogeneous nature of real world data. The model was trained on the PreSTS corpus comprising of 300 billion time points from various domains. Kairos achieves excellent forecasting performance on the GIFT-Eval benchmark, while having significantly fewer parameters compared to other models. Check the link for more information and follow me for regular data science content! 𝗞𝗮𝗶𝗿𝗼𝘀 𝗼𝗳𝗳𝗶𝗰𝗮𝗹 𝘄𝗲𝗯𝘀𝗶𝘁𝗲: https://lnkd.in/dtxjtQvK 𝗟𝗲𝗮𝗿𝗻 𝗠𝗟 𝗮𝗻𝗱 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴: https://lnkd.in/dyByK4F #datascience #python #deeplearning #forecasting

Standard LLM chains are great for straight lines, but real-world problems are messy. They need loops, second guesses, and human check-ins. LangGraph solves that problem. It’s the "pro mode" for LangChain. I just put together a quick post on how to get started with this framework with a simple code example. Here’s the TL;DR: ✅ Define State Early: Use TypedDict to give your agent a reliable memory. ✅ Conditional Edges: Let the LLM decide when to pivot or retry. ✅ Persistence: Never lose progress again with built-in checkpointers. ✅ Human-in-the-Loop: Add "pause" buttons for high-stakes actions. ✅ Keep Nodes Pure: Small, focused functions make debugging a breeze. Check out the full breakdown below! 👇 https://lnkd.in/e_T-bCaG #AI #DataScience #Python #LangChain #LangGraph #MachineLearning #LLM

🚀 Day 28 of Problem Solving Journey Today, I worked on an interesting problem: Group Anagrams 🔍 Problem Statement: Given an array of strings, group the anagrams together. Anagrams are words that have the same characters but arranged differently. 💡 Approaches I explored: ✅ Approach 1: Character Frequency Count (Optimized) Used a fixed-size array (26 letters) to count character occurrences Converted the count into a tuple to use as a dictionary key Achieved an efficient time complexity of O(n * k) ✅ Approach 2: Sorting Strings Sorted each string and used it as a key Simple and intuitive approach Time complexity: O(n * k log k) 📌 Key Learning: Understanding how hashing works with different representations (frequency vs sorted string) helps in optimizing solutions. ⚡ Takeaway: There are always multiple ways to solve a problem, but choosing the most efficient one makes a difference in real-world applications and interviews. 💻 Tech Used: Python | HashMap | Arrays #Day28 #ProblemSolving #Python #DataStructures #Algorithms #CodingJourney #100DaysOfCode

I have seen K-Means applied to data shaped like rings and crescents. The clusters were mathematically valid. They were completely meaningless. The algorithm was not wrong. The choice of algorithm was wrong. K-Means assumes spherical clusters of roughly equal size. Give it anything else and the math will produce results that look confident and explain nothing. Here is the actual decision: Clusters are roughly round and you know k. Use K-Means. Fast and interpretable. Data has irregular shapes or you need to detect outliers. Use DBSCAN. It labels noise as -1 and never forces points into clusters they do not belong in. You need to explore the structure without committing to k upfront. Use Hierarchical. Build the full tree and cut it wherever the dendrogram shows the largest gap. Measure your results with Silhouette score regardless of which algorithm you use. Swipe through for the code for all three. Which of these have you used on real data and what did you learn from it? #MachineLearning #Python #DataScience #UnsupervisedLearning