Charit Upadhyay
Adobe, Senior Site Reliability Engineer
San Francisco, California, United States
Actions
Charit Upadhyay is a Senior Site Reliability Engineer at Adobe, specializing in building scalable, secure, and high-performance cloud infrastructures. With extensive experience across Kubernetes, Terraform, observability, and security operations, he has led initiatives integrating AI into DevOps and cloud security workflows. Charit’s work focuses on applying emerging AI technologies to enhance operational efficiency, mitigate risks, and strengthen threat detection in complex systems. He is an active contributor to industry conferences, a reviewer for multiple technical committees, and a strong advocate for practical, real-world applications of AI in security and reliability engineering.
Links
Area of Expertise
Topics
Zero Trust for AI Systems: Securing What You Can’t Predict
Zero Trust has become a foundational security model, but it assumes predictable systems and well-defined behavior. AI systems challenge these assumptions by introducing dynamic decision-making, evolving inputs, and opaque execution paths.
This session explores how AI fundamentally breaks traditional Zero Trust boundaries and what needs to change.
We will cover:
New attack surfaces: prompt injection, data poisoning, model abuse
Why identity-based access control is insufficient for AI workflows
Applying Zero Trust principles to model inference and data flows
Securing AI pipelines across distributed cloud environments
Using real-world infrastructure patterns, we will demonstrate how to redesign Zero Trust architectures to account for autonomous systems.
Attendees will gain a practical framework to secure AI systems without sacrificing performance or scalability.
When AI Breaks Production: Designing SRE Guardrails for Autonomous Systems
As organizations rapidly integrate AI into production systems, a new class of failures is emerging — unpredictable, non-deterministic, and often invisible to traditional observability.
This talk explores real-world challenges faced when AI-driven systems intersect with production reliability, including:
Unbounded decision loops causing cascading failures
Silent data corruption from model drift
Latency spikes due to inference unpredictability
Security risks introduced by autonomous actions
Drawing from hands-on experience operating large-scale distributed systems, this session introduces a new reliability model for AI systems, where traditional SLOs and monitoring are no longer sufficient.
We will present a practical framework for:
Designing AI-aware SLOs (beyond latency & availability)
Implementing guardrails for autonomous systems
Building observability for non-deterministic behavior
Integrating Zero Trust principles into AI pipelines
Attendees will leave with actionable strategies to safely deploy AI in production without compromising reliability, security, or user trust.
Securing AI Pipelines: From Data Poisoning to Model Drift Detection
AI systems introduce new attack surfaces that traditional security models were not designed to handle. From data poisoning to prompt injection, and from supply-chain vulnerabilities to model drift, AI pipelines demand a new security mindset.
This session examines the evolving threat landscape around AI infrastructure and presents practical defense strategies, including:
Securing training and inference pipelines in cloud-native environments
Detecting and mitigating model drift in production
Protecting against data poisoning and prompt injection attacks
Applying zero-trust principles to AI systems
Integrating AI security into DevSecOps workflows
Designed for engineers and security leaders, this talk offers concrete architectural guidance for building AI systems that are not only intelligent—but resilient and secure.
The Hidden Reliability Crisis: Why Observability Fails in Modern Distributed Systems
Modern distributed systems generate more telemetry than ever before — yet outages are becoming harder to detect, diagnose, and prevent.
This talk explores why traditional observability approaches are failing in large-scale cloud environments, especially with microservices, Kubernetes, and multi-cloud architectures.
We will examine:
The gap between metrics and real system behavior
Why logs and traces miss cascading failures
Alert fatigue and the illusion of visibility
The impact of complex service dependencies
We introduce a practical approach to improving reliability:
Designing signal-driven observability instead of data-driven
Mapping service dependencies to failure patterns
Improving incident response with better context and correlation
This session is based on real-world experience managing production systems at scale and focuses on actionable improvements rather than tooling.
Operating AI Workloads in Production: Reliability Lessons from Real Systems
Most AI talks focus on building models. Few discuss what happens after deployment.
As organizations move AI systems from experimentation to mission-critical production environments, new reliability challenges emerge: GPU resource contention, burst inference traffic, model rollout failures, cost volatility, observability blind spots, and security risks across data pipelines.
In this session, we’ll explore practical, real-world lessons from operating AI workloads at scale in cloud-native environments. Attendees will learn:
How to design resilient infrastructure for AI inference and training workloads
Kubernetes strategies for scaling GPU-based systems
Blue/green and canary deployments for models
Observability patterns for AI systems (latency, drift, cost monitoring)
Securing AI pipelines without slowing innovation
This talk bridges the gap between ML innovation and production reliability. Attendees will leave with actionable architectural patterns for running AI systems safely, securely, and at scale.
LLMs in DevOps: Practical Automation Beyond Chatbots
AI in DevOps is often reduced to chatbots and code completion tools. But large language models can be used far beyond simple assistance—they can actively reduce operational risk and improve system resilience.
This session explores real-world applications of LLMs within DevOps and SRE workflows, including:
AI-assisted incident summarization and RCA drafting
Automated log analysis using embeddings
Infrastructure-as-Code generation and validation
Risk detection in CI/CD pipelines before production rollout
Intelligent runbook generation and policy enforcement
Rather than theoretical possibilities, this talk focuses on practical implementations and lessons learned when integrating AI into production reliability engineering.
Attendees will gain a framework for responsibly deploying LLM-powered automation inside their DevOps ecosystems—without introducing new operational or security risks.
From Reactive SRE to Predictive Reliability: Building Self-Healing Cloud Systems
Modern cloud systems are operating at a scale where traditional reactive SRE practices are no longer sufficient. Teams are overwhelmed by alerts, delayed incident response, and increasing system complexity across distributed and multi-cloud environments.
In this session, I will share how we evolved reliability engineering from reactive troubleshooting into a predictive and automated system. Using real-world production examples, we will explore how to detect failures before they happen and build systems that can recover automatically.
We will cover practical approaches to:
Designing self-healing systems using automation and observability
Reducing alert fatigue with intelligent signal correlation
Automating incident response and recovery workflows
Handling large-scale reliability challenges in Kubernetes and cloud-native environments
This talk focuses on real implementation strategies, lessons learned, and patterns that can be applied immediately to improve system reliability and reduce operational overhead.
When AI Agents Go Rogue - Securing Autonomous AI Systems Before They Act
Autonomous AI agents are no longer theoretical. They’re building workflows, calling APIs, writing code, and making decisions at scale. But with that power comes risk, new, emergent, and often unpredictable. As agent frameworks like AutoGPT, LangGraph, CrewAI, and custom orchestrators gain adoption, organizations must ask: What happens when your AI doesn’t just hallucinate but acts?
In this talk, Advait Patel, cloud security engineer and contributor to the Cloud Security Alliance’s AI Control Matrix, will unpack the risks associated with AI agents acting autonomously in production environments. Through real-world examples and red-team simulations, we’ll explore how agentic systems can be manipulated, coerced, or simply misaligned in ways that lead to security incidents, privacy violations, and cascading system failures.
Topics we’ll cover:
- How AI agents make decisions and where control is lost
- Prompt injection + tool usage = real-world lateral movement
- Over-permissive action spaces: API abuse, identity leaks, and shadow access paths
- Why traditional threat modeling fails for agentic workflows
- Techniques to sandbox, constrain, and monitor AI agents (function routers, policy-as-code, response filters)
- Logging and observability for “invisible” agent behavior
The attendees will walk away with:
- A framework to assess agentic AI security posture in your environment
- Examples of attack chains involving AI agents, cloud APIs, and dynamic plugin execution
- Architectural patterns to deploy secure-by-design agent frameworks in enterprise settings
- Recommendations for SOC teams on how to detect and respond to rogue agent behavior
This session is designed for CISOs, security architects, red teams, and AI product engineers who are exploring or deploying autonomous AI systems. If your LLM can act, it can be exploited, and this talk will show you how to defend against that future.
Can You Trust Your AI SOC Analyst? Testing the Limits of LLMs in Security Operations
LLMs are showing up in SOC tools, from log triage to incident summaries. But can we trust their outputs in critical workflows? This session explores the promises and pitfalls of using LLMs in security operations. We’ll evaluate real-world use cases like auto-generating detections, summarizing incidents, and helping with reverse engineering tasks. Through examples and benchmarks, we’ll explore where LLMs shine, where they hallucinate, and how to build secure, auditable pipelines around them. Attendees will leave with a framework to evaluate AI tools in the SOC, and a clear sense of when to automate, when to supervise, and when to just say no.
AGNTCon + MCPCon North America 2026 Sessionize Event Upcoming
CloudX 2026 Sessionize Event
AI DevSummit + DeveloperWeek Management 2026 Sessionize Event
Charit Upadhyay
Adobe, Senior Site Reliability Engineer
San Francisco, California, United States
Links
Actions
Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.
Jump to top