Vivek Singh

"Gaurav, how can I fine tune an LLM on user activity data?" My first mistake was to answer his question. "Pick a base model like Llama, and fine tune it using question and answer data." He didn't seem impressed, so I moved to the fundamentals. "What are you trying to do?" "I am trying to recommend users products based on their activity. Past browsing, searches and current trends." "You mean, you want to recommend stuff based on what they have bought previously?" "Yes." "So...two tower approach?" "What's that?" "I mean collaborative filtering and XGboost." "I don't understand." "Search for 'How to build a recommendation engine' on ChatGPT". "Okay, thanks..." I don't think he was convinced. For most technical problems, you don't need fine-tuning. You need clarity.

225

28 Comments

After reviewing 100s of Data Science resumes, I started noticing a pattern among top 1% candidates. The biggest differentiator wasn't technical skills. It was how they presented their story. Here are 3 things they do differently: ✅ They use the company’s language If the JD says “A/B testing to improve mobile features,” don’t just say “ran experiments.” Mirror the phrasing. It’s how recruiters spot alignment. ✅ They tweak their job titles to match the role If your title was “Operations Analyst” but your day-to-day looked like a Data Scientist’s, use that title instead. Align with what you actually did and what recruiters are looking for. ✅ They showcase portfolio projects like real experience Don’t just drop your GitHub link and call it a day. Add them to your experience section. Because that's what they are—experience. Most importantly: They don’t just apply. They position themselves for the role they want. 📌 I break down all 5 underrated strategies (and 3 common mistakes) in my latest article. (Find the full breakdown below 👇)

143

16 Comments

Myths that stops Data Scientists moving from SBC to PBC 🚫 1. PBCs only hire from top-tier colleges. I’ve seen people from tier-3 colleges land offers at Swiggy, Fractal, etc. Because strong portfolios and problem-solving skills always speak louder than a college tag. 2. “I’ll prepare for 6 months and then start applying.” Don’t make this mistake. If you’re aiming for a product company, build real-time use cases, keep your fundamentals strong, and apply in parallel. And this one’s my favorite: 3. “SBC experience isn’t valued in PBCs.” Not true at all. Experience is experience. What matters is the tech you’ve worked with, the projects you’ve built, and the impact you can prove. So if you’re in an SBC right now and dreaming of that switch: Stop doubting your background. Start sharpening your skills. Start building. Start applying. And a perfect place to start your journey would be Bosscoder Academy’s Data Science program. Check it here → bcalinks.com/iWliFcQ They’ve helped 2200+ professionals switch into Data & ML roles at top tech companies by offering: ✅Structured curriculum covering ML, Deep Learning & GenAI ✅Industry projects with real world datasets that build your portfolio ✅1:1 mentorship from Data Scientist at top product companies ✅Placement support to help you land your dream data role #myths #datascience #sponsoredpost Image Credit: Google Banana model 🙈

67

30 Comments

Every data scientist I know has the same ritual: - Save notebook every 2 minutes - Restart the kernel when it freezes - Pray nothing breaks during presentations Sound familiar? Here's the truth nobody talks about: data science and software development are 𝐍𝐎𝐓 the same. Yet we keep forcing ourselves into tools built for engineers, not data scientists. I've been testing Zerve AI, and honestly? It feels like cheating (the good kind). 𝐖𝐡𝐚𝐭 𝐦𝐚𝐤𝐞𝐬 𝐢𝐭 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭: → 𝐂𝐨𝐧𝐭𝐞𝐱𝐭-𝐀𝐰𝐚𝐫𝐞 𝐀𝐈 𝐀𝐠𝐞𝐧𝐭𝐬 – They understand your data, workflows, and compute environment. No more explaining the same thing 50 times. → 𝐕𝐢𝐬𝐮𝐚𝐥 𝐂𝐚𝐧𝐯𝐚𝐬 + 𝐍𝐨𝐭𝐞𝐛𝐨𝐨𝐤𝐬 – Drag, drop, connect blocks. Python, R, and SQL all in one place. When something breaks, you fix that block, not restart your entire Jupyter kernel. → 𝐁𝐮𝐢𝐥𝐭-𝐈𝐧 𝐎𝐫𝐜𝐡𝐞𝐬𝐭𝐫𝐚𝐭𝐢𝐨𝐧 – Parallel processing without the DevOps headache. Scale up, scale down, pay for what you use. The real game-changer? While you're debugging kernel issues in Jupyter, 𝐙𝐞𝐫𝐯𝐞 users are already deploying models. 𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐡𝐢𝐭 𝐦𝐞 𝐡𝐚𝐫𝐝𝐞𝐬𝐭: • Natural language to full workflows in minutes • Git integration that actually makes sense for data work • Real-time collaboration without the "who has the latest version?" Stop fighting with Jupyter environments that weren't built for modern data science. Your future self will thank you when you're shipping models instead of restarting kernels. ♻️ Share this with your data science team 𝐏.𝐒. I share job search tips and insights on data analytics & data science in my free newsletter. Join 18,000+ readers here → https://lnkd.in/dUfe4Ac6

218

79 Comments

🚩 You Can’t Grow in Data if You Don’t Learn This First Someone asked me this after my last post on business framing, and it’s such a good question. 1. Pause and Probe: When someone says “Can you pull customer data for Q2?” 
📍 Don’t open Jupyter Notebook immediately. Pause and ask:
→ What are they trying to decide?
→ What would make this analysis useful?
→ Are they asking the right question? 2. Learn to ask “Why?” and “Then what?” Someone says:
“We need a dashboard for marketing.” 📍 You say:
→ Why do you need it?
→ What decisions will it help you make?
→ What would you do differently after seeing it? You’re not being difficult. You’re helping. 3. Practice turning vague questions into specific ones: Business:
“Why are fewer people renewing their policy?” 📍 You learn to reframe: 
→ Is renewal rate dropping more for first-time customers vs returning ones? 
→ Did the drop begin after a pricing or communication change? This is what framing looks like. 4. Do this with any dataset you find: Go to Kaggle or anywhere. Pick a real-world dataset not dummy data. 
Before doing any analysis, do this: 📍 Ask yourself: 
→ If I was hired by this company, what problem would they want me to solve? → What would I ask them before writing code? 
→ What would be bad advice to give from this data? 📍This is how you practice the thinking. 5. Work backwards from the decision: 📍 Don’t start from the data. Start from the goal. → What’s the decision they want to make? 
→ What are they worried about? → What would “value” look like in this case? 📍That’s how you stop being a task-taker and start becoming a strategic thinker. → Business framing is not a tool. 
→ It won’t make your portfolio look fantastic. 
📍 But it will make you the analyst/scientists that gets invited into every meeting because you ask the right questions. 📍That’s how you move from executing tasks to influencing strategy. ♻️ Repost so others can learn #DataLiving #FavourIbude

474

41 Comments

👨‍💻 Data Science Interview question – Linear Regression Time Complexity A commonly asked question in data science interviews: 🧠 “What is the time complexity of linear regression?” ✅ A key follow-up should be: “Are we using the closed-form (normal equation) or gradient descent?” Because each has very different performance characteristics! 🔹 1. Closed-Form Solution (Normal Equation) Formula: θ = (XᵗX)⁻¹ Xᵗy Let: m = number of training examples n = number of features 🧮 Step-by-step time complexity breakdown: Matrix multiplication XᵗX (n × m with m × n): → Takes O(m × n × n) = O(n²m) 📝 (Because matrix multiplication of size A: m×n and B: n×p takes O(mnp) time) Matrix multiplication Xᵗy (n × m with m × 1): → O(nm) (smaller, absorbed into previous term) Solving (XᵗX)⁻¹ or linear system Aθ = b for A ∈ ℝⁿˣⁿ: → Typically solved using LU, Cholesky, or QR decomposition → Takes O(n³) time 📝 This is standard for matrix inversion or solving linear systems when A is n × n. 🔄 Total time complexity: O(n²m + n³) ✅ Use case: When the number of features (n) is small, and you want an exact, analytical solution. 🔹 2. Gradient Descent (Iterative Optimization) Minimizes the cost function by updating parameters in small steps. 🧮 Step-by-step per iteration: Compute Xθ: X is m × n, θ is n × 1 → result is m × 1 → O(mn) Compute residual (Xθ − y): → O(m) Compute gradient Xᵗ(Xθ − y): Xᵗ is n × m, residual is m × 1 → result is n × 1 → O(nm) 📌 Per iteration: O(mn) 📌 If K iterations: Total time complexity: O(Kmn) ✅ Use case: Scales well to large feature sets or massive datasets. Easily used in mini-batch or online learning setups. #DataScience #MachineLearning #LinearRegression #GradientDescent #DSInterview #MLAlgorithms #TimeComplexity #MatrixMath #BigDataML #MLEngineering

126

9 Comments

**So, why do you think profits dropped by 30%? 🤔** As analysts, we're often trained to jump straight into data dashboards, visualizations, and deep dives. But not every problem needs to start there. Thanks to a recent session led by Saurabh Kango Sir, I learned that when facing problems such as a sudden drop in profit or a spike in churn, we need to take a step back before diving into the numbers. That’s where structured thinking comes in — and frameworks help. --- 🔍 CRIED – For Product RCA Thinking: **Clarify the problem and what exactly is being measured. **Rule Out obvious errors like bugs, outliers, or tech glitches, economic crisis, recession 📊 Internal Diagnosis – TROPiCS: A Lens to Break Down Internal Factors **Time – Is it a sudden or gradual change? **Region – Is the trend isolated or global? **Other Metrics – What else is moving? Revenue, margins, engagement? **Platform – Is it app-specific? iOS? Android? **Cannibalization – Could a new feature be hurting another metric? **Segmentation – Which user groups are impacted? --- ###🌐 Don’t Ignore External Factors: Market trends, economic changes, competitor actions, and even geopolitical events can play a huge role. Last but not least is Data. Now that we have gone through every step above, we can deep dive into the data and find answers to our questions. Sometimes, the answer isn't in a pivot table — it's in the context. This shift in mindset has helped me reframe how I approach product and data problems — not every solution requires a technical approach. Sometimes, the best insights come from asking structured questions. #ProductAnalytics #RootCauseAnalysis #FrameworkThinking #Scaler #StructuredThinking #BusinessInsights #AnalyticsMindset

26

5 Comments