Vincent Caldeira
Leading Open Source Technology Innovation for a Sustainable Future
Singapore
Actions
Vincent Caldeira, Red Hat APAC CTO and Industry Visiting Scholar at Columbia University, drives tech strategy and emerging engineering. A Top 10 APAC CTO (2023) with 20+ years in finance IT, he is an authority on open source, cloud-native technologies and AI. Vincent holds leadership roles across the Linux Foundation (FINOS, LF AI & Data) and GSF, championing AI safety, sustainability, and digital sovereignty.
Badges
Area of Expertise
Topics
Trusting Agentic Workflows: Embedding Agent Evaluation into Fluxnova
Large Language Models enable workflows to handle tasks that were previously impossible to automate, but they also introduce new challenges around quality, governance, and trust. How do you confidently deploy agentic workflows when outcomes are inherently non-deterministic?
In this session, we demonstrate how Fluxnova's Agentic Subprocess capability can be combined with an open-source evaluation framework to create observable, governed, and continuously improving agentic workflows.
Using a financial services use case, we show how evaluation can be integrated throughout the workflow lifecycle: during development to validate behaviour before deployment, during execution through runtime evaluation and policy checks, and after execution using workflow traces to identify opportunities for improvement.
Attendees will see how Fluxnova combines deterministic workflow orchestration with adaptive AI capabilities, enabling organisations to blend traditional process control with agentic decision-making while maintaining transparency, auditability, and human oversight. The session includes practical implementation patterns and architecture guidance.
Distributed Agentic Systems & Workloads
This session gives an overview of the different initiatives towards establishing open standards for distributed agentic workload orchestration across CNCF, LFEdge & AAIF.
The content goes deeper into different aspects such as Infrastructure and platform requirements for large-scale agentic AI workloads, including deployment and operational considerations. The session covers runtime, orchestration, and distributed workload management including AIOps and Day 2 operations for distributed systems, and architectural patterns for agentic systems.
Cloud-Native Defense-in-Depth for Agentic AI: From AI Gateway to Kernel Space
The transition to production-grade agentic AI introduces critical security and governance challenges due to agent autonomy. This session details a multi-tier, cloud-native architecture combining the Kubernetes AI Gateway and OpenShell to establish defense-in-depth for autonomous AI workflows.
At the network and payload layer, the Kubernetes AI Gateway handles protocol translation (MCP/A2A), token rate-limiting, and payload inspection. At the execution layer, OpenShell provides kernel-level container sandboxing via eBPF, seccomp, and Landlock, paired with in-process OPA policy enforcement to prevent unauthorized syscalls and data access.
Building on CNCF cloud-native agentic standards and a 6-layer defense in depth framework for AI agents, we demonstrate how leveraging Kubernetes RBAC, SPIFFE workload identities, and isolated agent execution planes provides a scalable, enterprise-grade control plane for agentic AI.
Cloud Native Agentic Standards: A Progress Report from the CNCF AI TCG
As agentic AI expands across the cloud native ecosystem, a lack of uniform standardization and interoperability threatens secure enterprise scaling.
In this session, co-authors from the CNCF AI TCG provide an essential structural and operational update based on our recent foundational standards checklist for Kubernetes-driven workloads.We will break down the core pillars required for reliable multi-agent systems: communication protocols like MCP and A2A , identity and authorization frameworks including SPIFFE/SPIRE, granularity in time-series observability metrics , and flexible governance models designed to mitigate emergent AI behaviors.
Attendees will walk away with an agnostic view of cloud-native best practices to deploy securely, scale reliably, and ensure explainability across distributed agent networks.
An Open Blueprint for AI Compliance-as-Code
New AI and financial regulations such as NIST AI RMF, EU AI Act, DORA, force a hard question for the platform teams: how do we ship AI to production in regulated environments without stalling the release cycle?
This session, jointly presented by maintainers and contributors from OSCAL Compass (CNCF Sandbox), Sigstore Model Transparency (OpenSSF), and Kagenti (OSS), demonstrates end-to-end compliance automation for the AI lifecycle as a live walkthrough, from model release to agent operations.
We show how to digitize the EU AI Act into machine-readable OSCAL catalog, bring supply chain trust to ML models via Model Transparency signing, and enforce identity policies for agent-implemented controls through Kagenti. Posture assessment is automated using the right policy engine per control type, OPA, Kyverno, OCM, or CEL, with OSCAL Compass as the common standard for compliance artifacts representation.
Attendees will learn about our latest agentic architecture for compliance that scales in an open, composable, maintainable way, and where open specifications and policy open tooling meet in practice.
Eval-Driven Development: Mastering Agentic Tracing and Expert-Aligned Judges with MLflow
Building AI agents is easy, but ensuring their reliability is hard. Because agents operate in open-ended, non-deterministic environments, traditional Test-Driven Development (TDD) falls short. The solution is Eval-Driven Development (EDD), a paradigm that embeds continuous evaluation throughout the agent lifecycle.
This talk explores how to operationalize EDD. First, we examine the critical role of observability. Evaluating just a final output is insufficient for multi-step workflows. We show how OpenTelemetry (OTel) compatible tracing exposes internal reasoning and tool usage for granular debugging.
Next, we tackle scaling evaluation via LLM-as-a-Judge. Since uncalibrated judges often miss domain-specific nuances, we demonstrate aligning custom judges with human experts. Using open-source MLflow, attendees will learn to capture traces, collect expert feedback, and use alignment optimizers to create expert-aligned evaluators.
Stop Trusting a Black Box: The Economic Case for Open, Sovereign AI
As AI matures from a novelty into a strategic asset, 79% of organisations now prioritise sovereignty to mitigate vendor lock-in and secure critical data. Despite this, the market remains paralysed by a paradox: while open models have achieved performance parity with proprietary systems at a fraction of the cost, they remain massively under-utilised due to perceived friction and trust gaps.
This session dissects this market inefficiency, with insights from LF Research revealing how closed-source dominance is often driven by inertia rather than superior capability. We will explore how to operationalise true independence by rejecting "open-washing" in favour of rigorous frameworks that demand full model completeness: verifying everything from training data to weights. Join us to learn how to transition from renting black-box APIs to architecting a transparent, reproducible, and economically superior AI stack.
Stop Allocating GPUs, Start Delivering Intelligence:An Enterprise Blueprint for AI ROI on Kubernetes
For any enterprise, the high cost and chronic underutilization of GPUs is the single greatest threat to AI ROI. With one-third of cloud GPUs operating at less than 15% capacity, the key is to stop managing hardware and start delivering value.
This session presents a blueprint for transforming siloed GPU infrastructure into a centralized, high-yield "AI platform." We'll show why Kubernetes is the core economic engine that maximizes the return on your most expensive assets, not just a technical orchestrator. Learn how to:
Create a single, fungible GPU fabric to be shared across teams, boosting utilization.
Use intelligent autoscaling to match infrastructure spend to real-time demand.
Leverage vLLM and a new observability framework for distributed inference on kubernetes, llm-d, to manage performance SLOs (e.g., TTFT, TPS), giving you fine-grained control over token economics through tiered service offerings.
This blueprint is for platform engineers, MLOps leaders, and IT decision-makers aiming to justify AI spend and build an efficient foundation for innovation.
Operationalizing Agentic AI Safety & Evaluation for Multi-Agent Financial Systems
As financial AI shifts from passive models to autonomous agents, the industry faces a trust gap. Traditional "black box" validation is insufficient for systems executing complex workflows. This keynote explores the transition from MLOps to AgentOps, defining a standard for auditable Agentic AI where the decision process is as critical as the result.
We will dissect the FinSight Agent, a metacognitive FINOS Labs initiative built on LangGraph and MLflow, to demonstrate "Governance by Design." Aligning with the FINOS AI Evaluation Framework, we operationalize a "glass box" strategy. This moves beyond static benchmarks to implement trajectory tracing, where reasoning steps are audited against financial policies using LLM-as-a-Judge.
Finally, we cover system safety and supply chain security. We demonstrate how proactive Red Teaming detects risks like market manipulation and regulatory evasion. We also explore ensuring model integrity via the Model Openness Framework and Sigstore, proving open collaboration is key to building safe, compliant financial infrastructure.
Graph-Grounded Agentic Retrieval for Multi-Stage Reasoning over XBRL Financial Disclosures
The financial industry has long struggled to bridge the gap between standardized XBRL data and the nuanced reasoning needed for deep analysis. Traditional retrieval methods often treat complex disclosures as flat text, losing critical semantic relationships between line items, footnotes, and temporal periods.
This session introduces Graph-Grounded Agentic Retrieval. Using open-source tools like docling and docling-graph, we transform raw XBRL filings into hierarchical knowledge graphs that mirror the inherent structure of financial reporting. We also offer a data-driven demonstration of why graph-based approaches are superior for AI agents. Finally, as part of the FINOS AI Evaluation and Benchmarking stream, we use a rigorous framework to map the "reasoning trajectory" of agents navigating these graphs. Benchmarking against datasets like FinDER and FinAgentBench, we provide empirical evidence that grounding agents in document structure significantly reduces hallucinations and increases factual consistency over standard RAG.
The audience will leave with a blueprint for a verifiable, high-precision retrieval architecture for regulated financial content.
From Lab to Life: Practical AI System Evaluation
Agentic AI systems are a significant evolution from single-model GenAI Chatbots, but their dynamic and unpredictable nature in the real world introduces significant operational, reputational, and financial risks for enterprises. This "reality gap" is a critical blind spot that static, pre-deployment benchmarks like MMLU—with their fixed datasets—fail to address.
We propose a practical approach inspired by the framework suggested by the University of Michigan in their paper: "Evaluation Framework for AI Systems in the Wild".
The authors’ advocacy for holistic frameworks that integrate performance, fairness, and ethics can be seen as a foundation for a risk-adjusted evaluation. Their suggested use of continuous, outcome-oriented methods that combine human and automated assessments while also being transparent can increase trust among stakeholders.
We will break down the principles of the framework and provide practical, actionable approaches for a risk-adjusted evaluation using the best of open-source technologies. We will explore how to apply these evaluation methods throughout the entire AI system development lifecycle, from inception to continuous, real-world monitoring.
Engineering an Open-Source AML Detective: Local Graph Reasoning with SLMs and Edge AI
Financial institutions face a conflict between AML detection needs and data sovereignty. Cloud-based LLMs often struggle with the non-linear topologies of financial crime. This session presents a paradigm shift: an "AML Detective" agent for native graph reasoning on the edge, built with a 100% open-source stack.
We detail fine-tuning a 30B Mixture-of-Experts (MoE) model on the IBM AML Synthetic dataset. We show how Polars manages high-throughput transactions and NetworkX enables deterministic structural analysis. We explain our pivot from Reinforcement Learning (GRPO) to stable, process-driven Direct Preference Optimization (DPO) using Unsloth for memory-efficient training on local hardware. Using MLflow for trajectory tracking, our hybrid-judge system increased detection success by 10% while reducing investigation steps in half and ensuring sensitive data remains local. Attendees will learn to orchestrate these tools to build privacy-compliant, metacognitive agents bridging graph theory and generative AI.
Decoding the open-source blueprint for India's sovereign AI future
As India transitions from being the world’s largest consumer of open-source software to a leading creator of digital public infrastructure, a new mandate has emerged: Sovereign AI. To achieve true digital autonomy, Indian enterprises and public institutions must build AI ecosystems that protect data residency, reflect local context, and avoid vendor lock-in. But how do we practically build this sovereign stack?
This session provides an end-to-end technical blueprint for building enterprise-grade sovereign AI infrastructure entirely on open-source technologies. Using a homegrown, production-ready AI coding assistant as a practical case study, we will deconstruct the architectural layers required for AI independence. We will explore how to orchestrate scalable infrastructure with OpenStack, abstract complex multi-vendor GPU environments using Kubernetes, and deploy high-throughput inference for open-weight models using vLLM.
Beyond the architecture, we will discuss how Indian firms can adopt this open-source stack to implement highly secure, air-gapped environments, protect intellectual property, and empower local engineering talent to shift from consuming global AI to building it.
Beyond Monolithic AI: Cloud-Native Patterns for Dynamic Model Selection and Semantic Routing
The era of the "one-size-fits-all" LLM is ending. We are shifting toward Compound AI Systems—complex meshes where the goal isn't just to query a model, but to dynamically select the best model for the specific task at hand. This shift creates a massive opportunity for cloud-native architectures: how do you govern non-deterministic routing at scale?
This session breaks down the infrastructure required to move from monolithic agents to multi-model orchestration. We will demonstrate how to implement Semantic Routing within an AI Gateway to act as a traffic controller, instantly analyzing user intent to route queries to the most capable (or cost-effective) model. You will learn patterns for "supervisor" workflows, where lightweight models handle routing and heavyweight models handle self-correction. Join us to discover how to build controlled AI systems on Kubernetes, ensuring your agents are not just powerful, but precise, effectively governed, and fundamentally safer.
Architecting Secure Agentic Workflows on Kubernetes: A Financial Sector Case Study
As organizations move beyond basic LLM integrations toward autonomous agentic workflows, the infrastructure required to support these systems grows increasingly complex. Running multi-agent architectures in production introduces unique challenges around tool discovery, secure access, and traffic routing.
This session explores how to leverage Kubernetes and cloud-native abstractions to develop, test, and deploy AI agents at scale. Using a reference architecture leveraging the Kagenti project, we will demonstrate how to construct a secure, scalable agentic environment. The talk covers unifying tool access via an MCP Gateway, enforcing zero-trust workload identity with SPIFFE/SPIRE, and standardizing inter-agent communication.
To illustrate these concepts, we will walk through a real-world financial use case: an autonomous agent that securely accesses market data and executes simulated transactions using financial tools over MCP, demonstrating end-to-end cloud-native deployment.
Applied GenAI in Action: A Shared Framework for Financial AI Evals and Benchmarking
"Can this system be trusted?" is the hardest question to answer for GenAI in finance. General-purpose benchmarks, often limited to the model (like MMLU) fail to capture the nuance of regulated tasks, forcing institutions to build proprietary evaluation silos. This session unveils the current effort within the FINOS AI Evaluation & Benchmarking workstream, as part of our Applied GenAI initiative. We will explore how the community is mapping real-world financial use cases directly to rigorous evaluation metrics.
In particular, the presentation will feature a technical overview of current state-of-the-art evaluation techniques for multi-agent systems leveraging the FinSight Agent, an open-source reference implementation for analyzing corporate earnings calls. This use-case allows us to demonstrate how to execute consistent, reproducible tests using open datasets and synthetic data pipelines.
Discover how your firm can stop guessing and start measuring AI performance against an industry-standard baseline.
AI Agents & Platform Engineering: Efficiency Boost or New Source of Trouble?
As application developers accelerate their output with AI-assisted coding tools, platform engineers face mounting pressure to keep pace. Can nondeterministic AI agents help bridge this gap, or are they a troublesome new source of complexity and unpredictability? What common challenges cause these AI initiatives to stall, and what does a “minimum viable” platform foundation for success actually look like? How do teams evaluate agent effectiveness and manage the cost implications of AI in production? How can platform engineers build trust in nondeterministic systems, and how does the human–agent collaboration model differ from traditional team dynamics?
This vendor-neutral KubeCon panel brings together platform engineering leaders, project creators, and technical experts to share practical insights on the real-world impact of AI agents in platform engineering, offering proven patterns for defining golden metrics, avoiding common pitfalls.
Sovereign AI in Action: Advancing India’s Digital Future with Open Source Infrastructure
India’s pursuit of Sovereign AI—the ability to build and run AI systems within national borders—is fast becoming a strategic imperative. This session explores how open source principles and infrastructure are accelerating India’s journey toward digital self-reliance, highlighting collaboration between local cloud providers and the open source community.
Discover how a leading Indian datacenter and cloud provider is building a sovereign cloud platform that supports AI aligned with India’s regulatory, cultural, and linguistic landscape. With scalable, GPU-powered infrastructure and open frameworks, this platform empowers enterprises to train and deploy AI models securely and responsibly.
We’ll also showcase community-driven approaches to model customization that reduce barriers for domain experts and enable more relevant, safer AI. Learn how open tooling supports transparent governance, data residency, and innovation across sectors—from finance to healthcare.
Join us to see how open ecosystems are shaping India’s inclusive, self-reliant AI future.
Operationalizing Openness: Standardizing AI Model Supply Chains with the Model Openness Framework
As AI systems proliferate, ensuring transparency, trust, and traceability across the model supply chain has become a critical challenge. The Model Openness Framework (MOF), developed by LF AI & Data and Generative AI Commons, offers a standardized classification system to evaluate the completeness and openness of AI models across 17 key components—from architecture to evaluation code and documentation. This talk will explore how the MOF addresses model "openwashing" and supply chain risk by establishing clear standards for licensing and disclosure. We will demonstrate how enterprises can operationalize MOF compliance using open source tools like OCI-based model packaging, model signing, and automated documentation pipelines. Attendees will gain practical insights into aligning with emerging governance requirements and building trustworthy, reproducible AI systems through open collaboration.
Multi-Layered Guardrails for Cloud-Native AI: Enforcing Compliance and Safety at Scale
As AI-powered cloud-native applications evolve, ensuring trust, compliance, and robustness requires dynamic governance mechanisms that operate seamlessly across distributed environments. This keynote introduces a multi-layered cloud-native framework that enforces AI guardrails at three critical stages: pre-processing (input validation), inference (real-time bias mitigation), and post-inference (output validation).
By leveraging Kubernetes orchestration, Istio service mesh, and knowledge graphs, the framework enables scalable AI governance that integrates multi-agent coordination, real-time intervention, and traceability to ensure AI decisions remain transparent, auditable, and aligned with compliance requirements.
Attendees will gain insights into cloud-native AI governance patterns, practical deployment strategies, and the role of multi-agent oversight in ensuring compliant, production-ready AI workflows within Kubernetes environments.
From Containers to Cognitive Agents: Open Source Foundations for Enterprise-Grade AI Systems
As AI evolves from isolated model development to system-level deployments, enterprise AI engineers face increasing complexity in tooling, governance, and operational workflows. This session explores how open source technologies—Podman Desktop, RamaLama, and Llama Stack—can streamline the design and delivery of secure, scalable, and reproducible agentic AI systems. Attendees will follow an end-to-end user journey: starting from secure container-based local experimentation, advancing through modular multi-agent RAG workflows, and culminating in OCI-compliant model packaging and Kubernetes-native deployment. We’ll dive into practical strategies for integrating model governance via the LF AI & Data Model Openness Framework, achieving robust supply chain security, and enabling observability across the AI lifecycle. Whether you're building internal AI platforms or modernizing MLOps, this session reveals how open tooling empowers teams to confidently operationalize next-gen AI systems at enterprise scale.
Scaling AI Responsibly: Building Ethical, Sustainable, and Cloud-Native AI Systems
Panel Discussion - As AI continues to reshape industries, organizations face mounting pressure to scale AI systems responsibly while addressing challenges in efficiency, sustainability, and trust. This panel convenes leading experts to discuss how cloud-native technologies and CNCF projects are paving the way for scalable, ethical, and resource-efficient AI. Attendees will gain actionable insights into optimizing AI workflows, reducing environmental impact, and ensuring transparency in AI decision-making. From leveraging open-source tools to implementing cost-effective and ethical AI practices, this session will equip you with the knowledge to build AI systems that are both innovative and responsible. Discover how to harness the power of cloud-native ecosystems to drive AI transformation without compromising on sustainability or trust.
AI/ML engineers and data scientists looking to scale AI systems in cloud-native environments.
OS-Climate: A Data-Driven Open Source Approach to Climate-Aligned Finance Investing
LF OS-Climate (OS-C) is a breakthrough initiative creating a transparently governed public utility of open data and open source tools for climate-aligned finance investing, business, and regulation. This session explores the launch of a global open-source collaboration to create the most potent tool for data-driven decisions in transition strategies, investments, new technologies, and policies. We will present OS-C's comprehensive technical infrastructure and methodologies, fostering the development of open data products that address the challenges of climate transition. A critical focus will be the Data Mesh architecture, pivotal for federating diverse climate data sources and achieving Net Zero objectives. Additionally, we'll introduce a new open data product, essential for financial institutions to disclose quantitatively the degree of alignment/non-alignment to Paris goals of the $87 Trillion in sovereign bonds making up about a third of all pension fund and insurance investments globally, a vital step for the Net-Zero Asset Owner Alliance (AOA) and Glasgow Financial Alliance for Net Zero (GFANZ) members towards realigning commitments to Net Zero.
Trust in Green: Towards a Cloud Native Approach for Building Sustainable and Reliable Enterprise AI
As organisations increasingly integrate AI solutions, the demand for environmentally sustainable practices within this space has never been more critical. This presentation delves into the collaborative effort between the Cloud Native Computing Foundation (CNCF) AI WG and the TAG Environmental Sustainability to define a repeatable design approach aimed at fostering sustainable AI in cloud-native environments. Our discussion will outline the crucial considerations in such approach, including efficient management of compute resources, storage optimisation, and advanced networking solutions. Attendees will gain insights into the lifecycle of AI/ML deployments, from inception through operation, emphasising resilience, scalability, and resource efficiency. By highlighting innovative "green" strategies, this session will provide actionable best practices and recommendations, alongside a forward-looking perspective on future trends and research directions in sustainable AI.
Green AI in Cloud-Native Ecosystems: Strategies for Sustainability and Efficiency
The rapid proliferation of AI is increasing focus on the environmental costs associated with large-scale model training and deployment. As cloud-native technologies form the backbone of modern AI systems, the Cloud Native Computing Foundation (CNCF) is spearheading efforts to balance AI innovation with sustainability. This session will provide an overview of the CNCF effort to identify key areas, techniques, and best practices for energy-efficient AI in cloud-native environments. Attendees will gain insights into a newly developed taxonomy that categorises remediation patterns and sustainable practices across AI lifecycle phases, deployment environments, and personas.
We will also explore real-world applications and discuss reference architectures that provide means to optimise resource use, such as GPU slicing for inference efficiency, power capping during training, and carbon-aware scheduling, while maintaining performance and scalability.
OS-Climate and Unity Catalog: Pioneering Open Source ESG Data Sharing
In the rapidly evolving landscape of ESG data, financial institutions face the dual challenges of data fragmentation and security. This session will explore OS-Climate's innovative use of open source technologies, including Apache Iceberg, Apache Spark and the Unity Catalog, the industry's only universal catalog for data and AI, to enable secure and efficient ESG data sharing across financial institutions. Through this session, participants will gain insights into the strategic benefits of leveraging open source tools for ESG data management, including enhanced data security, governance, and interoperability. In particular we will demonstrate how Unity Catalog Unity's multimodal interface supporting various data formats and engines supports compatibility across diverse data ecosystems to facilitate seamless integration and governance of data assets, allowing financial institutions to not only meet regulatory compliance but also to harness ESG data for strategic advantage.
ONE Summit 2026 Sessionize Event Upcoming
Open Source Summit Japan + Automotive Linux Summit + Embedded Linux Conference Asia 2026 Sessionize Event Upcoming
Cloud Native AI Summit - EUROPE 2026 Sessionize Event Upcoming
CNCF-Hosted Co-located Events North America 2026 Sessionize Event Upcoming
Open Source in Finance Forum New York 2026 Sessionize Event Upcoming
KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 Sessionize Event
Open Source Summit Korea 2026 Sessionize Event
KubeCon + CloudNativeCon Japan 2026 Sessionize Event
Open Source in Finance Forum London 2026 Sessionize Event
KubeCon + CloudNativeCon India 2026 Sessionize Event
Open Source Summit India 2026 Sessionize Event
Open Source in Finance Forum Toronto 2026 Sessionize Event
KubeCon + CloudNativeCon Europe 2026 Sessionize Event
Open Source Summit Europe 2025 Sessionize Event
KubeCon + CloudNativeCon India 2025 Sessionize Event
Open Source Summit India 2025 Sessionize Event
KubeCon + CloudNativeCon Japan 2025 Sessionize Event
KubeCon + CloudNativeCon Europe 2025 Sessionize Event
KubeCon + CloudNativeCon India 2024 Sessionize Event
Open Source in Finance Forum New York 2024 Sessionize Event
Open Source Finance Forum London 2024 Sessionize Event
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