Rajeev Ranjan
Edison, New Jersey, United States
276 followers
275 connections
View mutual connections with Rajeev
Rajeev can introduce you to 10+ people at Barclays Investment Bank
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
View mutual connections with Rajeev
or
New to LinkedIn? Join now
By clicking Continue to join or sign in, you agree to LinkedIn’s User Agreement, Privacy Policy, and Cookie Policy.
Activity
276 followers
-
Rajeev Ranjan reposted thisRajeev Ranjan reposted thisMost engineers set up 1 layer out of 4. Then wonder why Claude feels useless. Here's the cheatsheet changed how I use Claude Code. Sharing it because it should change yours, too. Most people set up Claude Code like this: Run /init → get a generic CLAUDE.md → start prompting. Claude has no idea how your project works. So it hallucinates structure. Misses context. Again. Here's what actually matters: ▪️ CLAUDE.md = Claude's persistent memory. Loaded every session. ▪️ Add your tech stack, architecture, build commands. ▪️ Add gotchas Claude can't infer on its own. ▪️ Keep each file under 200 lines. Subfolder files append context. ▪️ Never overwrite parent context. Here's (exactly) how to set it up: 1. Install Claude Code → Requires Node 18+ → Run: curl -fsSL https://lnkd.in/d6x7eeDK | bash → cd your-project → claude → /init → It scans your codebase. Creates a starter memory file. 2. Build your CLAUDE.md properly → What: tech stack, directory map, architecture → Why: purpose of each module, design decisions → How: build/test/lint commands, workflows, gotchas 3. Add Skills (the real superpower) → Skills = markdown guides Claude auto-invokes → Project skill: .claude/skills/<name>/SKILL.md → Description field is critical for auto-activation → Use for: code-review, testing patterns, commit messages 4. Set up Hooks → Hooks = deterministic callbacks. PreToolUse / PostToolUse / Notification → Exit code 0 = allow. Code 2 = block. → Use for: security scripts, auto-lint, safety gates 5. Daily Workflow → cd project && claude → Shift + Tab + Tab → Plan Mode → Describe feature intent → Shift + Tab → Auto Accept → /compact → Esc Esc → rewind → Commit frequently. Start new session per feature. The 4-layer architecture behind it all: L1 – CLAUDE.md: persistent context and rules L2 – Skills: auto-invoked knowledge packs L3 – Hooks: safety gates and automation L4 – Agents: subagents with their own context Most developers use Layer 1 and wonder why it feels basic. Set up all 4. One afternoon. Then you'll understand why engineers switched. _________________________________ Want to go deeper on AI tools like this? Here you can find FREE Guides and Courses on AI workflows, agentic systems, and practical implementation. → https://lnkd.in/d6XFUESR ♻️ Repost to help your team stop winging Claude Code. Image credits: Brij Kishore Pandey #ClaudeCode #Engineering #DeveloperTools #AI #Productivity #Anthropic
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisIncredibly proud to watch my daughter graduate from MIT with her Master’s degree few days back! 🎓✨ Seeing her hard work, resilience, and passion culminate in this milestone has been the ultimate joy I can't wait to see the impact she will make next.
-
Rajeev Ranjan liked thisRajeev Ranjan liked this𝐌𝐨𝐬𝐭 𝐭𝐞𝐚𝐦𝐬 𝐭𝐚𝐥𝐤 𝐚𝐛𝐨𝐮𝐭 𝐑𝐀𝐆. Very few understand there is more than one way to build it. Wrong architecture. Poor retrieval. Higher costs. Slower responses. Frustrated users. The real question is not "Should we use RAG?" It is "Which RAG pattern fits our use case?" Here are the ones that matter right now 👇 → 🏗️ Standard RAG - The foundation. Document retrieval and question answering. → 🤖 Agentic RAG - Adds reasoning, planning, and tool usage for complex workflows. → 🕸️ Graph RAG - Connects relationships across entities and knowledge networks. → 🔀 Fusion RAG - Combines multiple retrieval strategies to improve relevance. → 🔁 Multi-Pass RAG - Iterative retrieval for deeper context discovery. → 💬 Interactive RAG - Refines results through user interaction and feedback. → 📈 Feedback-Based RAG - Continuously improves retrieval quality over time. → ⚡ Auto RAG - Automates retrieval and optimisation decisions. → 🧠 REALM - Integrates retrieval directly into model reasoning. The organisations winning with GenAI are not just deploying RAG. They are selecting the right retrieval architecture for the problem they are solving. Access to AI is no longer the advantage. How effectively your AI retrieves, reasons, and acts on knowledge is. 🎯 Which RAG architecture do you think will dominate enterprise AI in the next two years? 👇 Follow Raghavendra Bagalkoti for more on AI 🤖 × Cloud ☁️ × Capital Markets 📊 If you want to go deeper on AI and cloud concepts — built a free practice platform TechSkill. AWS, Azure, GCP, AI certifications & many more. Worth bookmarking if you are serious about getting certified. 👉 techskill.ai
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisEveryone's using Claude Code. Almost nobody understands what's happening under the hood. It feels like magic - you type, it codes, tests pass. But behind that simplicity is one of the most sophisticated agent architectures ever shipped to developers. Here's how Claude Code actually works 👇 🔹 𝗜𝗻𝗽𝘂𝘁 𝗟𝗮𝘆𝗲𝗿 User interface (CLI, IDE, CI/CD), session manager (resume, fork, persist), and a permission gate. Nothing runs without approval. 🔹 𝗞𝗻𝗼𝘄𝗹𝗲𝗱𝗴𝗲 𝗟𝗮𝘆𝗲𝗿 Skill registry, task graph, context compressor, and a memory store for cross-session persistence. This is how it "remembers." 🔹 𝗧𝗵𝗲 𝗠𝗮𝘀𝘁𝗲𝗿 𝗔𝗴𝗲𝗻𝘁 𝗟𝗼𝗼𝗽 Perception → Action → Observation. The heartbeat of the entire system. 🔹 𝗘𝘅𝗲𝗰𝘂𝘁𝗶𝗼𝗻 𝗟𝗮𝘆𝗲𝗿 Typed tool dispatch, streaming parallel runtime, and a prompt cache that cuts cost to ~10%. 🔹 𝗢𝗯𝘀𝗲𝗿𝘃𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝗟𝗮𝘆𝗲𝗿 Event bus + background executor - lifecycle hooks and interception baked in. 🔹 𝗠𝘂𝗹𝘁𝗶-𝗔𝗴𝗲𝗻𝘁 𝗟𝗮𝘆𝗲𝗿 Subagent spawning, teammate mailboxes, FSM protocol, autonomous board, and worktree isolation for zero-conflict parallel work. Claude Code isn't a chatbot that writes code. It's a full agent operating system. Understand the architecture, and you'll use it 10x better. Save this. Share with your eng team. #ClaudeCode #AI #AgenticAI #SoftwareEngineering #DevTools
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisMost people can’t explain the difference between LLM, RAG, AI Agents, and MCP. Here’s the simplest breakdown you’ll find: 🧠 LLM = The Brain → Reasons. Generates. Thinks. → But it’s frozen in time. No live data. Can hallucinate. 📚 RAG = Brain + Books → Plugs the LLM into your docs and databases → Retrieves context BEFORE answering → More reliable. Less hallucination. 🤲 AI Agents = Brain + Hands → Doesn’t just answer — it ACTS → Uses tools, memory, and multi-step planning → Automates workflows end to end 🔌 MCP = The Nervous System → Connects everything together → Standard protocol for tools, APIs, and data sources → The backbone that makes enterprise AI actually work The mental model: LLM thinks → RAG knows → Agents act → MCP connects Save this. Share it with someone who’s still confused. (Credit: Amit Kumar for the original analogy framework) Follow me for more on AI 🤖 × Cloud ☁️ × Capital Markets 📊 #GenerativeAI #AIAgents #PromptEngineering #CapitalMarkets #ArtificialIntelligence #Cloud
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisFrom Uncertainty to Quantum Insight: The Journey of Prof. Amit Agarwal Prof. Amit Agarwal is a theoretical physicist working at the interface of quantum transport, many-body physics, and quantum materials. Currently a professor at Indian Institute of Technology, Kanpur and a recipient of the 2025 Shanti Swarup Bhatnagar Prize, his work focuses on uncovering fundamental phenomena in quantum materials and topological systems, often with an eye toward experimental relevance. At IIT Kanpur, Prof. Agarwal has taken an interdisciplinary approach, collaborating with engineering departments and connecting fundamental theory with experimental platforms. His current research focuses on quantum transport and band-geometry-driven phenomena in quantum materials, with increasing emphasis on making experimentally testable predictions with real-world relevance. In recognition of his contributions, he was awarded the 2025 Rashtriya Vigyan Puraskar: Vigyan Yuva – Shanti Swarup Bhatnagar Prize, one of Indias highest honours in science and technology. Reflecting on his journey, Prof. Agarwal notes that while he continues to value the elegance of theoretical work, he is increasingly motivated to ensure that his research connects meaningfully with experiments and applications. Beyond academia, his interests remain simple. “I like to play chess sometimes,” he says. “My life largely revolves around family, research, and I immensely enjoy teaching. These are where my focus is." Read the full article here https://lnkd.in/gwFtNpYP Dr Vanita Srivastava | Ministry of Education, Government of India, New Delhi #IITKFaculty #FacultySpotlight #Spotlight #AcademicJourney #IITKanpur #IITK #PhysicsFrom Uncertainty to Quantum Insight: The Journey of Prof. Amit AgarwalFrom Uncertainty to Quantum Insight: The Journey of Prof. Amit AgarwalIndian Institute of Technology, Kanpur
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisStop using Claude like a search engine. The ones getting real results treat it like a system. Here's the 8-file setup that changes everything: 1. about-me.md: Your identity Claude knows nothing about you by default. This file fixes that. Your role, company, and current priorities Your background and what matters this quarter Everything Claude needs before it starts any task 2. voice-profile.md: Your taste DNA Stop getting generic AI responses. This file makes Claude sound like you. How you think, write, and see the world Your beliefs, writing mechanics, aesthetic preferences What you love and what makes you cringe 3. anti-ai-writing-style.md: Your boundaries The most underrated file in any Claude setup. Everything Claude should NEVER sound like Ban words like 'delve,' 'synergy,' 'utilize' No corporate speak, no overly enthusiastic tones 4. The Cowork Folder: Your 4-folder system Structure Claude's workspace so it always knows where everything lives. ABOUT ME/ → identity + writing rules PROJECTS/ → briefs, drafts, references TEMPLATES/ → finished work as patterns CLAUDE OUTPUTS/ → where Claude delivers 5. Global Instructions: Your persistent rules Set it once. Never type it again. Always reads your ABOUT ME/ before starting Settings → Cowork → Edit Global Instructions 6. The One Prompt: How you start every chat One prompt structure that forces Claude to ask YOU questions first. I want to [TASK] so that [SUCCESS CRITERIA] Read folder → Ask questions → Refine → Execute 7. Connectors: Claude inside your tools No more copy-pasting between apps. Plug into Slack, Google Drive, Notion, Gmail Settings → Connectors → Browse → Add 8. Plugins: Instant skill packs Pre-built workflows for specific jobs — ready to install in seconds. Marketing, Data, Sales plugins available Cowork → Customize → Browse Plugins → Install Which of these 8 are you missing from your setup? 👇 -------------------------------------------- Subscribe to our newsletter for daily AI news: https://lnkd.in/dzYgVBjF -------------------------------------------- #Claude #AIProductivity #Cheatsheet #AITools
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisMost engineers set up 1 layer out of 4. Then wonder why Claude feels useless. Here's the cheatsheet changed how I use Claude Code. Sharing it because it should change yours, too. Most people set up Claude Code like this: Run /init → get a generic CLAUDE.md → start prompting. Claude has no idea how your project works. So it hallucinates structure. Misses context. Again. Here's what actually matters: ▪️ CLAUDE.md = Claude's persistent memory. Loaded every session. ▪️ Add your tech stack, architecture, build commands. ▪️ Add gotchas Claude can't infer on its own. ▪️ Keep each file under 200 lines. Subfolder files append context. ▪️ Never overwrite parent context. Here's (exactly) how to set it up: 1. Install Claude Code → Requires Node 18+ → Run: curl -fsSL https://lnkd.in/d6x7eeDK | bash → cd your-project → claude → /init → It scans your codebase. Creates a starter memory file. 2. Build your CLAUDE.md properly → What: tech stack, directory map, architecture → Why: purpose of each module, design decisions → How: build/test/lint commands, workflows, gotchas 3. Add Skills (the real superpower) → Skills = markdown guides Claude auto-invokes → Project skill: .claude/skills/<name>/SKILL.md → Description field is critical for auto-activation → Use for: code-review, testing patterns, commit messages 4. Set up Hooks → Hooks = deterministic callbacks. PreToolUse / PostToolUse / Notification → Exit code 0 = allow. Code 2 = block. → Use for: security scripts, auto-lint, safety gates 5. Daily Workflow → cd project && claude → Shift + Tab + Tab → Plan Mode → Describe feature intent → Shift + Tab → Auto Accept → /compact → Esc Esc → rewind → Commit frequently. Start new session per feature. The 4-layer architecture behind it all: L1 – CLAUDE.md: persistent context and rules L2 – Skills: auto-invoked knowledge packs L3 – Hooks: safety gates and automation L4 – Agents: subagents with their own context Most developers use Layer 1 and wonder why it feels basic. Set up all 4. One afternoon. Then you'll understand why engineers switched. _________________________________ Want to go deeper on AI tools like this? Here you can find FREE Guides and Courses on AI workflows, agentic systems, and practical implementation. → https://lnkd.in/d6XFUESR ♻️ Repost to help your team stop winging Claude Code. Image credits: Brij Kishore Pandey #ClaudeCode #Engineering #DeveloperTools #AI #Productivity #Anthropic
-
Rajeev Ranjan liked thisRajeev Ranjan liked thisI’m grateful to have completed Year 3 of the Securities Industry Institute (SII) at The Wharton School of Executive Education—an important milestone in my journey of continuous learning and leadership development. This year was particularly special as we celebrated the 75th anniversary of the Securities Industry Institute at Wharton, a program that has helped develop generations of leaders across the financial services industry. The SII, organized by the SIFMA and delivered in partnership with The Wharton School of Executive Education, is one of the premier leadership development programs for senior professionals in financial services. The program brings together leaders from across the industry to deepen strategic thinking, strengthen leadership capabilities, and explore the rapidly evolving landscape of global markets, regulation, innovation, and client expectations. Completing the third year of this program reflects a personal commitment to investing in growth, learning from world-class faculty, and engaging with an incredible cohort of industry leaders. I’m deeply grateful to Fidelity Investments for nominating and sponsoring me for this program and for its continued investment in developing leaders across the organization. I also want to thank my fellow SII classmates and the outstanding faculty—including, like Professor Jeremy J. Siegel, Peter Conti-Brown, Lauren Hirshon, Dr. Roch Parayre, and many others—whose insights, discussions, and collaboration made this experience truly exceptional. #FidelityAssociate
Experience
Education
Licenses & Certifications
View Rajeev’s full profile
-
See who you know in common
-
Get introduced
-
Contact Rajeev directly
Other similar profiles
Explore more posts
-
Rajath Sakkari K
IG Group • 1K followers
MCP - a open protocol for AI (LLM) to talk to other tools / systems / applications. Naive, but don’t judge me, I was judging AI capabilities with me being Naive - So I asked Claude AI to create an MCP - An MCP the AI itself will use - Gave a website address to which MCP had to be created (with public APIs) - It Generated extensive, detailed code - I dwelled on it, fine tuned for my preference / requirements, it went ahead and did a bit more than I asked for. Website was analysed in minutes (amusing and scary) - I asked it to package it as executable package - I had configured file system MCP - So, Claude leveraged that and created a python project in filesystem and created Project with proper files and folder structure - Here comes Claude code (CC) - This python project of MCP server, was not in a running state - Asked CC to make the code runnable (of course I was in control (manual loop) of the changes it was making) - MCP server to run in local system (MacOs) needed some changes in trivial python code, CC made the changes - Interesting side part, during this whole process, it was also creating documentation, to install server, to run server and the MCP config for other tools - At times I had to say update the documentation - OOPS !! Code got changed, but what was changed ? - I asked CC to make this project version controlled. Oh my Devaa (God), it converted the project into a Git repo (local) - Created git related files, Auto committed all files. All I had to do was, say yes, go ahead - Well, we got side tracked, lets get back to MCP - So, now I asked CC to run the server (lazy to copy the command and run it in next tab terminal) - Obviously.. It had issues, and obviously it auto corrected the code - It wrote test cases, wrote test start script, tested it, committed the code and then - now it tells me, 'All fixed, Want to run?', Howdhu maga howdhu, run (yes dude yes, run) - MCP server started. I updated MCP configuration and restarted Claude Desktop - pssssttt….. curios? - :( Error: MCP server couldn’t be started - Again me being extremely (purposefully) lazy, told CC, you are being dumb, fix the startup error - I guess, I hurt its Ego. - It asked permissions for MCP server log file. whaaaat really ?, it wants to read log files now, wow. - It test ran with its test scripts, read the log, analysed like a nerd, updated the test files and code, reran until it was fixed and updated the documentation and also committed the code - What can I say, I was watching Christopher Nolan movie level events. - Finally, MCP worked, I tested few things, results were as expected. It ain’t over, my architectural / Engineering hormones kicked in. How about containerisation ? How about productionising it ? How about Security & Authentication ? How about Performance & Scalability ? How about Operational Readiness ? How about AI Security, Gaurdrails ? Stay tuned, I will be back 🙂
14
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content