Articles by Mohit
-
Best Practices for Secure Code Review
Best Practices for Secure Code Review
To prevent our designed application from security threats we must focus more on the underlying code. When an…
Activity
3K followers
-
Mohit Kumar posted thisMcKinsey charges substantial fees to guide companies on digital transformation. And yet, their own AI platform was reportedly compromised using vulnerabilities like SQL Injection and broken access control, issues that have been around for decades. This wasn’t some sophisticated zero-day attack. It was basic security hygiene… missed. What’s more interesting (and concerning) is how it happened. The attacker wasn’t a human manually probing systems. It was an AI agent, able to identify and chain vulnerabilities on its own. That’s the real shift. The attack surface is evolving rapidly with AI. But in many cases, the fundamentals of security still aren’t keeping up. No matter how advanced the technology gets, ignoring the basics can still bring everything down. And now, with AI accelerating both innovation and attacks, those gaps can be exploited faster than ever. So the question is: Who’s assessing your AI systems today?
-
Mohit Kumar reposted thisMohit Kumar reposted thisGoogle released massive learning resources for AI Agents 10+ Code Samples, Whitepapers, Hands-on Projects and much more... Google recently finished wrapping up their latest 5-Days intensive Kaggle AI Agent course where they gave us bunch of resources to learn and practise building AI Agents. Day 1: Introduction to Agents Understand AI agent fundamentals: core architecture, taxonomy of capabilities, and how agents differ from LLMs. Learn to build agents that perceive, plan, act autonomously. Whitepaper: https://lnkd.in/grYivvCW Code Resource 1: https://lnkd.in/gcDruAx7 Code Resource 2: https://lnkd.in/gmYMFkDX Day 2: Agent Tools & Interoperability with MCP Learn how agents interact with external systems through tools and APIs. Understand MCP architecture, tool design best practices, and human-in-loop approval workflows. Whitepaper: https://lnkd.in/gnU9yqqW Code Resource 1: https://lnkd.in/g5ZQHGzg Code Resource 2: https://lnkd.in/gPsBR3k7 Day 3: Context Engineering: Sessions & Memory Discover how agents maintain context across conversations. Understand Sessions for immediate state and Memory for persistent learning across interactions. Whitepaper: https://lnkd.in/g9WztfuP Code Resource 1: https://lnkd.in/g4mQEtPE Code Resource 2: https://lnkd.in/gRtJFvi5 Day 4: Agent Quality Master observability foundations: Logs, Traces, Metrics. Learn evaluation frameworks including LLM-as-a-Judge and Human-in-the-Loop for production reliability. Whitepaper: https://lnkd.in/g-SAMSpV Code Resource 1: https://lnkd.in/gJxMN46g Code Resource 2: https://lnkd.in/gR_QKbrS Day 5: Prototype to Production Understand deployment lifecycle and scaling strategies. Learn Agent2Agent Protocol for system-wide coordination and production readiness on Vertex AI. Whitepaper: https://lnkd.in/gnUAscjM Code Resource 1: https://lnkd.in/gnikixYA Code Resource 2: https://lnkd.in/gUBBWhxS Note: Learn the prerequisites before running these solutions as they require a few unique configs and API keys. (Check comments) If you want to understand AI agent concepts deeper, my free newsletter breaks down everything you need to know: https://lnkd.in/g5-QgaX4 💾 Save ➞ 👍 React ➞ ♻️ Share & follow for everything AI Agents
-
Mohit Kumar reposted thisMohit Kumar reposted thisMoving from "Can we build it?" to "Is it safe?" 🛡️ Everyone has seen the RAG Developer's Stack, but as we move from POC to production, the conversation must shift to security. You can't just deploy and pray. I’ve mapped out the AI Security Stack—the essential layers needed to protect your LLM architecture. Here is the breakdown of the defense layers: 🔹 Model Supply Chain: Scanning proprietary and open models (Hidden Layer, Protect AI). 🔹 Hardening & Guardrails: preventing jailbreaks and hallucinations (NeMo, Lakera). 🔹 Adversarial Testing: Red teaming before you ship (Red Team Toolkit, Robust Intelligence). 🔹 Real-Time Monitoring: Catching anomalies in the wild (Splunk, Datadog). 🔹 Data Governance: Ensuring privacy compliance (Privacera, BigID). Building the AI is only half the battle. Securing it is the war. Which layer is your team prioritizing right now? 👇 #AISecurity #LLMOps #DevSecOps #GenerativeAI #Cybersecurity #RAG
-
Mohit Kumar shared thisA great opportunity to get hands-on experience architecting AI agents using Gemini and Antigravity. My experience using Antigravity has been nothing short of amazing.Mohit Kumar shared thisI’m excited to share that #GoogleCloud Labs is returning to Singapore! 🇸🇬 On 25 March, we’re hosting a full day of technical workshops at the Google Developer Space. This year, we’re moving beyond chatbots to the next era of Autonomous Agents. Here is what I’m most looking forward to: 🚀 Gemini 3.0 & ADK: Building autonomous agents with real business logic. 📚 Knowledge-Driven RAG: Ensuring hallucination-free results. ⚡ AntiGravity: Next-gen #GenAI developer tools. 🛠️ Gemini Enterprise: Scaling AI with a no-code approach. Seats are strictly limited for this hands-on experience. Secure your spot here: https://lnkd.in/gHYvdRFZ #GoogleCloudLabs
-
Mohit Kumar shared thisApplication security just took a major leap in shifting towards “Secure by AI” dev cycles. Anthropic dropped “Claude Code Security” yesterday, and it's already sending shockwaves through the trillion-dollar software & cybersecurity ecosystem. Built on the freshly-released Claude Opus 4.6, it scans entire codebases with deep contextual reasoning. It uncovers novel, high-severity vulnerabilities that traditional SAST, SCA, and even seasoned human reviewers miss for years. -Its testing revealed over 500 long-hidden bugs in mature open-source projects that had endured decades of expert audits. It goes further: assigns severity + confidence scores, proposes precise patch suggestions, and runs multi-stage verification to slash false positives. On the flip side, it gives power to the bad actors too. As this model democratizes the vuln discovery by lowering the barriers for good and bad actors. This feels like the moment "secure by design" evolving into "secure by AI." Anthropic article: https://lnkd.in/gUaVUaP9 #AISecurity #Cybersecurity #DevSecOps #Anthropic #SecureByAI #ApplicationSecurity #BusinessRisk #InnovationMaking frontier cybersecurity capabilities available to defendersMaking frontier cybersecurity capabilities available to defenders
-
Mohit Kumar reposted thisMohit Kumar reposted thisIf you’re the kind of cybersecurity professional who thrives on leading security operations, driving incident response, outsmarting attackers, shaping detections, and mentoring analysts — this role has your name on it. Mediacorp is hiring an Assistant Lead for Cyber Defence & Resilience to guide analysts, turn threat intelligence into action, and safeguard platforms that millions rely on every single day. This is more than just protecting systems — it’s about protecting trust. Click the link to apply or happy to pass along referrals.. #cybersecurity #SOC #incidentresponse #threatintelligence #detectionengineering #mediasecurity
-
Mohit Kumar shared this37% of global cloud Infra using AI GPU is vulnerable to this CVE. NVIDIAScape (CVE-2025-23266) is more than a vulnerability - it’s a reminder and precaution to Cyber leaders and Boardroom. We’ve all been talking about AI risks in terms of model integrity, adversarial inputs, and data misuse. But the disclosure of NVIDIAScape is a reminder that the real fragility often lies underneath the AI stack, in the infrastructure we assume is safe. A three-line exploit can break out of NVIDIA AI containers and compromise entire GPU hosts. In a multi-tenant cloud world, that means cross-tenant breaches, leaked models, and disrupted AI services at scale. The key takeaway: AI risk is not new risk — it’s amplified risk. Speed matters: Patching and mitigating is urgent so that window for attackers is short. Board conversations must evolve. Securing AI is not only about the algorithm rather it’s about the ecosystem it runs on. Our role is to safeguard not just innovation, but the infrastructure that makes it possible. #AI #RiskManagement #CloudSecurity #CVE202523266 https://lnkd.in/g4Ybk7eNNVIDIAScape - NVIDIA AI Vulnerability (CVE-2025-23266) | Wiz BlogNVIDIAScape - NVIDIA AI Vulnerability (CVE-2025-23266) | Wiz Blog
-
Mohit Kumar shared this🌟Ranked as the 2nd highest scorer in Singapore! 🌟I'm thrilled to announce that I've successfully earned my Certified Information Security Manager (CISM) certification! 🎓💼 Throughout this journey, I've delved deep into critical domains such as information security governance, risk management, and incident management, equipping myself with the knowledge and skills necessary to navigate today's complex cybersecurity landscape. I want to express my sincere gratitude to my mentors, colleagues, and the vibrant cybersecurity community for their unwavering support and guidance. If you're considering pursuing the CISM certification or have any questions about my experience, feel free to reach out. #CISM #Cybersecurity #Certification #ProfessionalDevelopment #InfoSec https://lnkd.in/gkR8s3wnCertified Information Security Manager® (CISM) was issued by ISACA to Mohit Kumar.Certified Information Security Manager® (CISM) was issued by ISACA to Mohit Kumar.
-
Mohit Kumar posted thisHard selling vs Endorsements- Opting for hard selling can be effective for early career professionals or companies. However, it is endorsements that truly amplify one's career or brand. Therefore, the focus should lean towards garnering multiple endorsements, even if it requires an investment of time, effort, and a steadfast commitment to long-term goals. Capitalise on awards and accolades as it gives people who do not know us an idea of where we stand compared to others. #career #endorsement #hardselling
Mohit Kumar commented on a post
2w
Congratulations! 🎉
-
Mohit Kumar liked thisMohit Kumar liked thisAfter all the noise around Project Glasswing and the usual “this is the end of cybersecurity” takes flying around, I wanted to take a real step back. I have read a lot of doom, a lot of fear and a lot of people acting like the only conclusion here is panic. Yes, Anthropic’s Mythos capability is a genuine powerful shift. Finding vulnerabilities at that scale, including bugs hiding in production code for decades, is a big deal. The discovery side of the problem has clearly moved on. But the real story for me is what comes next. This does not kill cybersecurity. It exposes where the real work has always been. Prioritisation. Ownership. Software QA. Remediation capacity. Change velocity. Engineering discipline. That is where most organisations still struggle. Not because nobody cares, but because real estates are messy, old and full of systems nobody wants to touch on a late Friday afternoon. So rather than add to the doom spiral, I wrote my take on the more optimistic side of this. Where I disagree with some of the commentary. What this actually changes. What it does not change. And where I think the industry has a real opportunity to get better. Better QA before code ships. Smarter release discipline. AI assisted remediation where it genuinely helps. And more honest, intelligence led prioritisation instead of pretending every red dot is the same fire. There is still a lot of good work to do. There are still good people doing it. And there is a credible path forward here if we stop admiring the problem and start fixing the pipeline around it. That is my view on the state of play and what to do next. Link in the comments. #CyberSecurity #CISO #VulnerabilityManagement #AI #ProjectGlasswing #SecurityLeadership
-
Mohit Kumar liked thisMohit Kumar liked thisWhat most people get wrong about organizational politics and the security program. Creating and running an enterprise wide security program requires you to navigate, and benefit from, organization politics. This takes influencing skill and executive presence. These aren’t simply innate skills. They can be developed like any other skill. Navigate the organization. https://lnkd.in/eCEfhz2s
-
Mohit Kumar liked thisMohit Kumar liked thisGoogle released massive learning resources for AI Agents 10+ Code Samples, Whitepapers, Hands-on Projects and much more... Google recently finished wrapping up their latest 5-Days intensive Kaggle AI Agent course where they gave us bunch of resources to learn and practise building AI Agents. Day 1: Introduction to Agents Understand AI agent fundamentals: core architecture, taxonomy of capabilities, and how agents differ from LLMs. Learn to build agents that perceive, plan, act autonomously. Whitepaper: https://lnkd.in/grYivvCW Code Resource 1: https://lnkd.in/gcDruAx7 Code Resource 2: https://lnkd.in/gmYMFkDX Day 2: Agent Tools & Interoperability with MCP Learn how agents interact with external systems through tools and APIs. Understand MCP architecture, tool design best practices, and human-in-loop approval workflows. Whitepaper: https://lnkd.in/gnU9yqqW Code Resource 1: https://lnkd.in/g5ZQHGzg Code Resource 2: https://lnkd.in/gPsBR3k7 Day 3: Context Engineering: Sessions & Memory Discover how agents maintain context across conversations. Understand Sessions for immediate state and Memory for persistent learning across interactions. Whitepaper: https://lnkd.in/g9WztfuP Code Resource 1: https://lnkd.in/g4mQEtPE Code Resource 2: https://lnkd.in/gRtJFvi5 Day 4: Agent Quality Master observability foundations: Logs, Traces, Metrics. Learn evaluation frameworks including LLM-as-a-Judge and Human-in-the-Loop for production reliability. Whitepaper: https://lnkd.in/g-SAMSpV Code Resource 1: https://lnkd.in/gJxMN46g Code Resource 2: https://lnkd.in/gR_QKbrS Day 5: Prototype to Production Understand deployment lifecycle and scaling strategies. Learn Agent2Agent Protocol for system-wide coordination and production readiness on Vertex AI. Whitepaper: https://lnkd.in/gnUAscjM Code Resource 1: https://lnkd.in/gnikixYA Code Resource 2: https://lnkd.in/gUBBWhxS Note: Learn the prerequisites before running these solutions as they require a few unique configs and API keys. (Check comments) If you want to understand AI agent concepts deeper, my free newsletter breaks down everything you need to know: https://lnkd.in/g5-QgaX4 💾 Save ➞ 👍 React ➞ ♻️ Share & follow for everything AI Agents
Experience & Education
-
Mediacorp Pte Ltd
View Mohit’s full experience
See their title, tenure and more.
or
Licenses & Certifications
Volunteer Experience
Languages
-
English
-
-
German
-
-
Hindi
-
Recommendations received
1 person has recommended Mohit
Join now to viewOther similar profiles
Explore more posts
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More