Engineering Director | "People" first - not "resources" | Drive Large Scale Distributed Platforms | AI/ML - DeepLearning | Data Engineering/Big Data | GCP Architect | Financial Risk | Learner

Seasoned Engineering Director, currently driving Digital Acquisitions at American Express, with more than 20+ years of experience leading large-scale applications in fintechs. Proven track record in driving over $40 million tech investments per year and managing products impacting 20+ global markets, 10M+ users and 10,000+ servicing center reps. Expertise in AI/ML, GenAI, Data Engineering/Analytics and financial technologies (credit and fraud risk) using Java/Python/Go. Good experience in LLM training, data curation, and fine-tuning foundation models like Llama. Seeking engineering leadership
roles in driving exponential business growth.

Skills
• Technical Leadership Technical roadmap for AI/ML and GenAI and delivery of Large-Scale Distributed Applications
• Strategic Planning Partnering & Collaborating with C-Level Executives, Researchers for technical and product strategy.
• People Leadership Exemplar of organization culture, Building and mentoring cohesive global distributed teams.
• Influencer Managing key external customer technical relationships including Amazon, Lowes, Delta, Hilton.

✉ shail.bade@gmail.com

Shailendra Bade

AI agents are no longer living in laboratories. 
They're out in the wild, making decisions, writing code, and connecting with everything in sight. 
And guess what? 
The old security rules don't work anymore. ⚠️

Security baked in, not bolted on. 
We've all heard it. 
But now, with autonomous AI agents, it's not just a nice idea. 
It's survival. 🔒

Let's talk reality: Meta's LlamaFirewall is a wake-up call. 
It's not just another firewall. 
It's a signal that attackers are catching up to AI, and our playbook needs an overhaul. 🚨

Here's how the attack surface has changed: 
- Jailbreaking 2.0: Hackers don't just tweak outputs, they hijack agent goals. 
- Goal hijacking: Feed the agent the wrong input and you can reroute its entire mission. 
- Code injection at scale: If your AI writes code, every output is a security risk.

I've run my own POCs using layered defenses, inspired by this new approach. 
Real numbers, real results: 
- Input filtering blocked threats with over 97% accuracy. 
- Behavioral monitoring using lightweight models caught agents straying from intended goals. 
- Output validation checked every line before execution. ✅

The outcome? 
- Successful attacks dropped from nearly 18% to under 2% in controlled tests. 
- Small, efficient models (86M parameters) outperformed their heavyweight cousins for security tasks. 
- False alarms stayed below 2.5% and didn't slow down the systems.

Here's the uncomfortable truth: 
If your AI agent can search the web, write code, and act on its own, the entire system becomes an attack surface. 
The security fence moves with every new feature you ship.

The organizations that win in the autonomous AI era? 
They're the ones building security into the DNA of their AI from the very beginning. 
Not via quick patches. 
Not by "adding a firewall later." 
But by rethinking how AI is designed, deployed, and monitored. 🛡️

Leaders, the time for afterthought security is over. 
If your plan starts with "we'll add protection later," you're already behind.

Engineering Director | Gen AI/RAG |AI/ML & LLM Systems at scale | From Financial Fraud Detection to Personalization | Data Engineering & Privacy

taplio

The Battle

Shailendra Bade

Kumar Subramaniam

Followers

Posts

Engagements

Likes

Performance ScoreShailendra Bade's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Shailendra Bade

Kumar Subramaniam

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Shailendra Bade's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreShailendra Bade's LinkedIn Performance