Divanshu Garg

While interviewing candidates for GenAI roles, I’ve noticed a common pattern, almost everyone claims to have worked on GenAI projects, especially chatbots/RAG/ Q&A using LLMs. However, many struggle with basic NLP and ML fundamentals. Basic questions like: What embedding techniques have you used? What’s the role of encoder-decoder architecture? Why did you choose an LLM over a simpler model? ...often go unanswered. GenAI is powerful, but it's not a one-stop solution for every problem. Many problems are better solved with traditional NLP and ML techniques. Here’s my advice to the candidates appearing for GenAI interviews: 1. Don’t skip the basics- Learn traditional NLP: tokenization, embeddings (Word2Vec, GloVe, FastText, BERT), attention, seq2seq models, etc. 2. Understand classical ML- Some tasks are best solved with logistic regression or decision trees, not always an LLM. 3. Be clear on your GenAI project- If you list a GenAI project on your resume, know every step: data pipeline, model choice, fine-tuning, evaluation, deployment, limitations. 4. Learn when NOT to use GenAI- It’s not always the most efficient or cost-effective tool for the job. 5. Focus on depth- Real impact comes from understanding, not just using pre-trained APIs. Even after working on production-ready GenAI systems, we are still learning and evolving. Let’s stop chasing trends blindly and focus on building strong fundamentals. #Al #NLP #GenAl #MachineLearning #LLM #Interviewtips

1,406

110 Comments

A Redditor (𝘣𝘰𝘣𝘴𝘩𝘮𝘶𝘳𝘥𝘵) who secured a job at FAANG reveals the ultimate hacks for acing interviews at any company: 𝗦𝘁𝗲𝗽 𝟭: Skipping the Black Hole ▶️I never applied through company portals. They get thousands of applications, and ATS filters out most of them. ▶️Instead, I targeted engineers and hiring managers on LinkedIn and asked for referrals. ▶️I kept my messages short and to the point: "Hey [Name], I'm really interested in [Team/Company] and l'd love to apply. I have [X years] of experience in [Relevant Skill], and I think l'd be a great fit. Would you be open to referring me?" ▶️This got me multiple referrals in a week, and I went straight to recruiter screens instead of waiting in the void. 𝗦𝘁𝗲𝗽 𝟮: Only Studying What Actually Gets Asked ▶️Instead of grinding hundreds of LeetCode problems, I reverse-engineered the interview questions: ▶️I searched Glassdoor, Blind, and LeetCode discussion forums for recent questions from my target company. ▶️I found patterns-most companies ask the same 10-15 core problems repeatedly. ▶️Instead of solving 500 random problems, I studied: ▶️Top 30 questions per company (sorted by frequency) ▶️Patterns, not solutions (e.g., "Oh, this is just a sliding window problem with a twist.") ▶️Mock interviews on Pramp and with friends to get real-time feedback. ▶️Result? I was solving interview questions in under 10 minutes instead of struggling through brute-force solutions. 𝗦𝘁𝗲𝗽 𝟯: Finessing the Behavioral Interview ▶️FAANG behavioral rounds aren't about "personality" —they're looking for structured answers ▶️I prepped 5 stories using the STAR method (Situation, Task, Action, Result) and adapted them on the fly. ▶️The key? Always show impact with metrics. Instead of saying: "I helped optimize a backend service." I said: "I optimized the backend service, reducing latency by 40% and saving $500K in cloud costs." ▶️Biggest trick? If they ask about failure, always spin it into a win ("I learned X, and it led to Y success later"). 𝗦𝘁𝗲𝗽 𝟰: Exploiting the Hiring Process Loopholes ▶️I timed my interviews strategically-companies move faster when they know you have other offers. ▶️I sought out hiring events and "bar-raiser" systems (Amazon, for example, has bar-raisers who can override bad interviewers). ▶️I built relationships with my recruiter-they have power to push through borderline candidates and help with negotiations. 𝗦𝘁𝗲𝗽 𝟱: Offer and Negotiation Hacks ▶️Once I had one offer, I used it to pressure other companies to move faster. ▶️I acted slightly disinterested-companies chase candidates who seem in demand. ▶️I negotiated hard: ▶️"I love the opportunity, but my other offer is at $X-can you match or improve it?" ▶️"| was hoping for a higher base/signing bonus to align with market rates." ▶️Result? +$40K increase in total compensation. The 𝗘𝗻𝗱 𝗥𝗲𝘀𝘂𝗹𝘁? ▶️FAANG offer with $300K+ total comp ▶️Minimal time wasted on irrelevant prep ▶️Less stress, more control over the process

71

2 Comments

I often hear from early-career ML/DS folks who feel stuck or from folks who want to move in AI/ML. Here’s what I’d recommend based on my own hiring experience: 💻 Reframe your resume - One page only: Focus on the last 1–2 years of most relevant work. If >10 years, add second page. Make sure you can explain each and every project/ algo in detail. - Bullet points over paragraphs: “Built and deployed an anomaly detector in PyTorch on 1M records with 95% precision” beats “Experience with PyTorch.” - Group courses vs. projects: Put certificates (DeepLearning.AI, MLOps, GCP) in a small “Training” section; highlight actual projects and results first. This can also go into the skill section. Make sure whatever you put, you've actually worked on. 💻 Show production readiness - Recruiters want more than “I trained a model.” - Emphasize (learn if needed) end-to-end work: data pipelines, Docker containers, CI/CD, cloud deployments, monitoring. 💻 Build a portfolio of small wins - Pick 2–3 mini-projects that mimic real problems: train a neural network, RAG chatbot, data ETL pipeline, simple API serving a model. Perhaps share your experience on medium/blog. - Share code on GitHub with a clear README and hosted demo (e.g., Streamlit). Create PR requests to open source projects, if possible. 💻 Ace coding challenges - Practice timed Python/data-structure problems on LeetCode or CodeSignal. Can't emphasize how important and useful this is, purely for interview. - Practice again. Even when not interviewing. 💻 Network with purpose - Ask for 15-minute “resume reviews” or quick feedback on a project. - Engage on Discord communities (e.g., MLOps, DataTalks) — personal referrals carry weight. Landing that first ML role can feel like an uphill climb, but every resume tweak, project post, or friendly referral moves you closer. Keep iterating—and you’ll break through.

309

24 Comments

Don’t miss these essential skills for becoming a Senior Data Scientist. Over the years, I’ve applied to numerous companies and gone through a wide range of interview processes. Here’s what I realized: while interview styles differ across companies, certain patterns remain consistent. So instead of preparing randomly, focus your efforts on the right set of skills. Use this roadmap to guide your next big career move: 1. Foundation Stage • Math & Stats: Probability, Bayes’ theorem, descriptive stats, linear algebra, calculus basics • Python: Core syntax, NumPy, Pandas, Matplotlib/Seaborn, Scikit-learn, Jupyter notebooks • SQL: Joins, CTEs, window functions, complex queries • Data Wrangling: Missing values, feature engineering, outliers, transformations, EDA 2. Intermediate Level • Machine Learning: Regression, decision trees, SVMs, kNN, clustering, PCA, evaluation metrics • Databases: Advanced SQL, NoSQL (MongoDB/Cassandra), query optimization • Data Engineering: ETL pipelines, Airflow, APIs, data quality • Version Control: Git workflow, GitHub collaboration 3. Advanced Skills • Deep Learning: Neural networks, CNNs, RNNs, TensorFlow/PyTorch, transfer learning • NLP: Text preprocessing, embeddings, transformers, BERT • Big Data: Spark, Hadoop, distributed computing • Cloud & MLOps: AWS/GCP/Azure, Docker, MLflow, CI/CD • Tools: Streamlit/Flask, experiment tracking 4. Career Development • Portfolio: 3–5 end-to-end projects, Kaggle, GitHub • Business Skills: Problem framing, stakeholder communication, domain knowledge • Interview Prep: Core concepts, SQL/DSA challenges, system design, behavioral rounds Focus on one skill at a time. Mix theory with real projects. Revisit the basics often and engage with the data science community. Track your progress with milestones that matter. Balancing work, life, and learning is hard — especially with courses piling up. Choose what truly adds value. If you’re looking for structured, impactful learning — check out Bosscoder Academy https://bit.ly/4crzg5b They’ve helped 800+ professionals make successful career switches. As always, please do your own research before you enroll for any program #DataScience #AI #MachineLearning #GenerativeAI #CareerGrowth #Sponsored

241

13 Comments

The irony of hiring for AI roles in 2026. 🤖 📉 Today, I was screening resumes for a Senior Associate position. I opened a profile for an AI/ML specialist and saw this in the professional summary[ See image] Yes, you read that right. A direct copy-paste from ChatGPT, brackets and all. The Reality Check: AI is a tool to amplify your brilliance, not a replacement for your attention to detail. If you’re applying to a dream company, "careless" shouldn't be the first word that comes to a recruiter's mind. If you’re using LLMs to polish your CV, run this checklist first: 1. The "Bracket" Hunt: Did you remove the AI’s instructions? 2. The Logic Test: Does the flow sound like a human or a manual? 3. The Proof: Are your specific project wins actually in there? 4. The Authenticity Factor: Does this sound like you, or just a generic template? Don't let a "copy-paste" error stand between you and your next big role. 👇 Recruiters & Hiring Managers: What’s the wildest AI "hallucination" or mistake you’ve seen on a resume lately? #hiring #generativeai #careeradvice #techhiring

53

15 Comments

Been getting lots of questions about how to land an Applied Scientist / Research Scientist role — here’s what worked for me: 1️⃣ Master core coding patterns – Focus on Strings, Trees, Graphs, DP. Depth > quantity. 2️⃣ Know ML in breadth & depth – Classic ML + job-relevant deep learning topics. Understand the full ML pipeline — from large-scale training to real-world deployment. 3️⃣ Prepare behavioral answers like a scientist – Structure a few solid stories that map to company values. 🎯 Pro tip: Don’t wait for clarity — create it. Opportunities rarely arrive in perfect form. Apply boldly, explore broadly, and let momentum reveal your path. #appliedscientist #researchscientist #machinelearning #jobsearch #generativeai #careeradvice #datascience

195

3 Comments

AI won't replace analysts. But analysts who use AI as a thinking partner will replace those who don't. Three of the sharpest data people I know just launched a course to help you become that analyst. Sravya (Senior Manager, Data Science @ Grammarly), Hai Guan (Head of Data @ Ontra), and Shane Butler (Principal Data Scientist @ Ontra) created AI Analytics for Builders on Maven. 𝐖𝐡𝐚𝐭 𝐭𝐡𝐞 𝐜𝐨𝐮𝐫𝐬𝐞 𝐭𝐞𝐚𝐜𝐡𝐞𝐬: → Translate vague product questions into clear, testable hypotheses → Define primary, secondary, and guardrail metrics with AI → Break down KPI movements through segmentation and trend analysis → Pressure-test your assumptions and edge cases → Turn analysis into clear, actionable recommendations 𝐓𝐡𝐞 𝐝𝐞𝐚𝐥: 🔥 5 early bird spots at $180 (normally $1,200) 🎟️ Everyone else who applies gets 40% off ⏰ Applications close in 24 hours I've worked with these folks. They're sharp, practical, and actually care about teaching skills that stick. If you're in data, product, or analytics, this is worth a look. 👉 𝐀𝐩𝐩𝐥𝐲 𝐡𝐞𝐫𝐞: https://lnkd.in/g7AHwSYY ♻️ Repost to help someone in your network level up their analytics game

71

24 Comments

Data Science Interview Question: In a spam detection system, you observe high precision but low recall. How would you solve for this? First, lets understand whats happening - Precision = How many predicted spams are actually spam Recall = How many real spam messages are caught High precision but low recall often means the decision threshold is too high — the model only labels spam when it’s very confident. Most classifiers output a probability (e.g., 0 to 1 for spam). The default threshold is usually 0.5. So one solution would be - lower the threshold (e.g., to 0.3 or 0.2) to catch more spam, increasing recall (at the cost of some precision). Another way, is to choose a threshold using a valiadation data and selecting a metric that helps balance precision and recall. Example - Fβ Score: Allows you to weight recall more than precision (e.g., F2 if missing spam is worse than flagging non-spam). After choosing a metrics, you can test for range of threshold on validations datset. 𝗟𝗶𝗸𝗲 for more such content. 𝐒𝐮𝐛𝐬𝐜𝐫𝐢𝐛𝐞 to the FREE substack - https://lnkd.in/g5YDsjex and YT Channel - https://lnkd.in/gttKwJtd to land your next Data Science Role. 𝗙𝗼𝗹𝗹𝗼𝘄 Karun Thankachan for all things Data Science.

279

30 Comments

From Rejection to Realization: Why ML Engineers Need Gen AI and System Design Skills After clearing multiple interview rounds for a data scientist role, I faced unexpected questions in the final round: "How would you build a scalable AI system for millions of users?" "Can you enhance model performance using Gen AI?" Despite my ML expertise, I lacked the system design and Gen AI knowledge to answer confidently. That rejection made me realize a crucial gap. This is Why Gen AI and System Design Are Must-Have Skills Gen AI for Innovation: Enables creating synthetic data, generating insights, and building adaptive AI models. System Design for Scalability: Ensures real-time performance, efficient data pipelines, and fault-tolerant systems. Companies seek engineers who can design and implement AI systems that are scalable, cost-efficient, and impactful. With time I gained the confidence to design end-to-end AI solutions after upskilling. If you're an ML engineer aiming for growth, mastering Gen AI and System Design isn’t just a topic. Looking back, understanding Gen AI and System Design didn’t just help me crack interviews.It made me a better data scientist. If you’re preparing for data science interviews, I’d highly recommend taking these areas seriously. So to help you with this i would suggest Bosscoder Academy, a course designed for working professionals with mentors from MAANG companies. Check their complete course here: https://bit.ly/4c7P90M They have helped 800+ professionals switch to top companies. Get a ��𝐈𝐌𝐈𝐓𝐄𝐃 𝐓𝐈𝐌𝐄 𝐎𝐅𝐅𝐄𝐑 𝐨𝐟 25% 𝐃𝐈𝐒𝐂𝐎𝐔𝐍𝐓 on all their courses. #DataScientist #SQL #Python #CareerGrowth #InterviewPreparation

395

34 Comments

Chain-of-Thought prompts are on 🔥 right now, but that's only half the story of how we teach LLMs to think. So, how does it work? In this post, you will learn - 1. Why is data alone not enough for better reasoning? 2. How do LLMs reason through problems? 3. How does the reasoning toggle work to give different quality of responses? --- If you've liked the crisp content, then repost it to uplift your network ♻️ And follow Bhavishya Pandit to stay ahead in Generative AI! 🚀 👉 Join 1k+ professionals and leaders in my free newsletter, where I simplify cutting-edge AI without the jargon. Link in the comments! #llm #opensource #llama #huggingface #google #ibm #meta #openai #rag #python #datascience #machinelearning #ml #ai #artificialintelligence #nvidia #machinelearning #genai #generativeai #anthropic #claude

358

42 Comments

If you are targeting multiple roles while applying, make sure you have different resumes for each of them. If you apply for ML Engineer but 80% of your resume is for Full Stack development, then that tells recruiters nothing about your experience with ML, and forces them to put your application in the rejection pile. Make sure your project descriptions make sense, If you build some models, don't just mention "leading to 30% percent decrease in tickets". What does that mean? How are you getting tickets for a personal project? Some people have nice projects but they butcher them in their resumes because of the need to add quantifiable numbers to pass the ATS filter.

72

2 Comments

🧠 If you're building apps with LLMs, this paper is a must-read. Researchers at Microsoft and Salesforce recently released LLMs Get Lost in Multi-Turn Conversation — and the findings resonate with our experience at Vellum. They ran 200,000+ simulations across 15 top models, comparing performance on the same task in two modes: - Single-turn (user provides a well-specified prompt upfront) - Multi-turn (user reveals task requirements gradually — like real users do) The result? ✅ 90% avg accuracy in single-turn 💬 65% avg accuracy in multi-turn 🔻 -39% performance drop across the board 😬 Unreliability more than doubled Even the best models get lost when the task unfolds over multiple messages. They latch onto early assumptions, generate bloated answers, and fail to adapt when more info arrives. For application builders, this changes how we think about evaluation and reliability: - One-shot prompt benchmarks ≠ user reality - Multi-turn behavior needs to be a first-class test case - Agents and wrappers won’t fix everything — the underlying model still gets confused This paper validates something we've seen in the wild: the moment users interact conversationally, reliability tanks — unless you're deliberate about managing context, fallback strategies, and prompt structure. 📌 If you’re building on LLMs, read this. Test differently. Optimize for the real-world path, not the happy path.

170

12 Comments