Talks & Events

Talk at Google DeepMind on building trustable AI systems beyond simple chatbots.

Key points:

• Moving from conversational AI to agentic AI architecture
• Safety frameworks for production AI deployments
• The SAIF (Secure AI Framework) approach
• Trust and reliability in AI-driven decision making

Takeaway: Building trustable AI requires going beyond model accuracy to include safety frameworks, adversarial testing, and transparent decision-making processes.

Talk at BSides Seattle 2026 on architecting safe and trustworthy AI agents.

Key points:

• Adversarial robustness for multi-agent AI systems
• Defense frameworks against coordinated attacks
• Security architecture for production AI agents
• Multi-agent adversarial evaluation arenas
• Lessons from building AI systems at PayPal scale

Takeaway: As AI agents become more autonomous, adversarial testing and defense frameworks are essential — building safe AI requires thinking like an attacker.

Talk at Google I/O 2025 GDE Summit on bridging research and practice.

Key points:

• From academic research to production AI at PayPal
• The GDE program’s role in knowledge transfer
• Making cutting-edge research accessible to developers
• Real-world AI deployment challenges and solutions

Takeaway: The gap between research and production is bridgeable — GDEs play a key role in translating academic advances into practical guidance.

Talk at Google I/O GDE Summit 2025 on making academic research accessible to practitioners.

Key points:

• Translating PhD research into real-world products
• Lessons from building AI systems at PayPal
• Bridging the gap between academia and industry
• The GDE program as a bridge between research and practice

Takeaway: Research impact multiplies when academic findings are translated into accessible products and shared with the practitioner community.

Talk at Drexel University on AI red teaming with Gemini and Vertex AI.

Key points:

• AI red teaming methodology and best practices
• Using Google Gemini and Vertex AI for adversarial testing
• The SAIF (Secure AI Framework) approach to AI security
• Identifying and mitigating vulnerabilities in AI systems
• Real-world red teaming case studies

Takeaway: AI red teaming is essential before deployment — the SAIF framework provides a systematic approach to finding and fixing AI vulnerabilities.

Talk at KubeCon 2024 on supercharging GenAI inference.

Key points:

• Distributed inference with Ray for scaling LLM serving
• Kubernetes orchestration for ML workloads
• TPU acceleration for production AI inference
• Performance benchmarks and cost optimization
• Real-world deployment architecture

Takeaway: Combining Ray, Kubernetes, and TPUs provides a scalable, cost-effective infrastructure for production GenAI inference at enterprise scale.

Talk at Google Korea on LLM reasoning and agentic Gemma.

Key points:

• Building agentic AI systems with Google Gemma models
• Reasoning and planning capabilities of open-weight models
• Tool use and function calling with Gemma
• On-device vs cloud deployment trade-offs

Takeaway: Open-weight models like Gemma can power sophisticated agentic AI systems with proper prompting and tool integration.

Talk at Google Mountain View on LLMinABox, an on-device AI diary and concierge.

Key points:

• On-device LLM with Google Gemma for privacy
• Voice-driven interaction for personal AI
• Personalized diary and concierge without cloud data sharing
• Architecture for running LLMs locally on consumer devices

Takeaway: Privacy-preserving personal AI is possible today — Gemma runs locally, keeping your data on your device, never in the cloud.

Talk at GDG Houston on the state of WebXR.

Key points:

• Current capabilities of the WebXR API
• Building immersive experiences with A-Frame
• Google I/O announcements related to web technologies
• Future of mixed reality on the web

Takeaway: WebXR has matured significantly — now is the time to start building immersive web experiences.

Walkthrough of Google Gemini and Gemma models on cloud and on-device.

Key points:

• Gemini model capabilities on Google Cloud
• Gemma open-weight models for on-device deployment
• Cloud vs on-device trade-offs: latency, privacy, cost
• Practical use cases for each approach
• Live demos of both deployment modes

Takeaway: Choose Gemini on cloud for maximum capability and Gemma on-device for privacy and latency — both have their place in production AI.

Talk on building applications that interact with desktop, code, and files using LLMs.

Key points:

• RAG for interacting with local files, PDFs, and code
• Multimodal AI for understanding documents
• Using your own local LLM or Google Gemini
• Voice and text interaction in any language
• Privacy-preserving local AI processing

Takeaway: You can build a personal AI assistant that reads your files, understands your code, and responds in your language — all while keeping data private.

Talk at Google I/O Extended Edmonton 2024 on creating immersive art with AI.

Key points:

• Combining generative AI models with WebXR
• Interactive artistic experiences in the browser
• AI-powered creative tools for immersive media
• Live demo of AI-generated immersive art

Takeaway: Generative AI and WebXR together open new creative possibilities for artists and developers in the browser.

Talk at Stanford CS25 2024 on emergent abilities and scaling in LLMs.

Key points:

• How large language models develop unexpected capabilities as they scale
• Emergent abilities: capabilities not present in smaller models but appearing in larger ones
• Implications for AI safety and alignment
• Current research directions in LLM scaling

Takeaway: Understanding emergent abilities is critical for predicting and controlling LLM behavior as models continue to scale.

Talk at Google Cloud Research Innovator Showcase on using TPUs as cryptographic accelerators.

Key points:

• Leveraging AI hardware (TPUs/NPUs) for polynomial multiplication
• Accelerating Fully Homomorphic Encryption (FHE) and Zero-Knowledge Proofs (ZKPs)
• Adapting cryptographic operations to AI-centric architectures
• Performance evaluation and limitations

Takeaway: AI accelerators like TPUs can be repurposed for cryptographic computations, potentially accelerating privacy-preserving technologies like FHE and ZKPs.

Talk explaining what Google Developer Experts (GDE) are.

Key points:

• The GDE program: what it is and how to join
• GDE community and contributions
• How GDEs help the Google developer ecosystem
• Stories from GDEs around the world

Takeaway: GDEs are passionate developers who share their expertise with the community — the program is open to anyone with deep knowledge and a desire to help others.

Deep dive talk on LangChain as a Google Developer Expert.

Key points:

• LangChain architecture: chains, agents, memory, tools
• Building production LLM applications
• Agent patterns for tool use and reasoning
• Integration with vector stores for RAG

Takeaway: LangChain provides the building blocks for production LLM applications — chains for pipelines, agents for autonomous reasoning, and memory for context.

Talk at GDG Zagreb on the state of WebXR and building with AR/VR today.

Key points:

• State of WebXR API across browsers
• Building AR and VR experiences for the web
• Latest AR/VR technologies and tools
• Live demos of WebXR applications

Takeaway: AR/VR development for the web is accessible today with WebXR, enabling cross-platform immersive experiences without native SDKs.

GDG Zagreb Online Tech Talk covering Apache Kafka, Android PhotoPicker, and WebXR.

Key points:

• State of WebXR and what you can build today with AR/VR
• Cross-platform AR/VR development using web standards
• Adapting to device capabilities from mobile to VR headsets

Takeaway: The web platform supports immersive AR/VR experiences that work across all devices with a browser.

Talk on SolMover, a framework for smart contract code translation based on concepts.

Key points:

• Translating Solidity smart contracts to Move using LLMs
• Concept-based retrieval for low-resource languages
• Planning, coding, and debugging agents in an iterative process
• Substantial accuracy improvements over baseline models

Takeaway: Concept-based translation enables LLMs to generate code in languages they have never seen, bridging the gap for verifiable smart contract languages like Move.

Talk presenting UniTranslator, a multi-agent framework for safe code translation.

Key points:

• Re-imagining code translation as collaborative multi-LLM endeavor
• Natural Language Inference (NLI) grounding for translation safety
• Handling low-resource languages like Move
• Mitigating code artifacts and hallucinations through iterative feedback

Takeaway: Multi-agent collaboration with small LLMs can rival large monolithic models for code translation, especially for domain-specific and low-resource languages.

Talk at Rice University Seminar on mitigating hallucinations in AI-driven medical diagnosis.

Key points:

• Hallucination patterns in LLM-generated medical diagnoses
• Techniques to detect and reduce false medical outputs
• Safety considerations for AI in healthcare
• Evaluation frameworks for medical AI reliability

Takeaway: AI in healthcare requires extra vigilance — hallucination mitigation strategies are essential before deploying LLMs in medical diagnosis.

A conversation at Google I/O 2024 about how AI and LLMs are shaping the future of the web.

Key points:

• How agentic AI is transforming web applications
• On-device inference vs cloud-based AI trade-offs
• The intersection of AI with web technologies and standards
• Real-world AI deployment challenges at PayPal scale

Takeaway: The web platform is evolving to embrace AI natively, with on-device models and agentic frameworks becoming production-ready.

Talk at Berkeley RDI Research Meeting on SolMover for smart contract translation.

Key points:

• The challenge of generating Move smart contracts with limited training data
• Knowledge graph of Move concepts for retrieval-augmented generation
• Iterative refinement with planning, coding, and debugging agents
• Evaluation with open-source LLMs showing accuracy improvements

Takeaway: ConMover/SolMover addresses the low-resource code generation challenge by teaching LLMs concepts rather than syntax, enabling translation to new verifiable languages.

Talk at Berkeley RDI on using LLMs for safe code translation between programming languages.

Key points:

• The SolMover framework for concept-based code translation
• Translating smart contracts between Solidity and Move
• Multi-agent collaboration for safe code generation
• Handling low-resource and domain-specific languages

Takeaway: LLMs can translate code between languages they have never seen by learning concepts rather than syntax, enabling safe translation for domain-specific languages.

Talk at Columbia University on privacy-aware zero-knowledge login.

Key points:

• Zero-knowledge proof authentication
• OAuth2 with Passkey integration
• Privacy-preserving login without sharing credentials
• Modern authentication best practices

Takeaway: Zero-knowledge login with Passkeys provides stronger security than passwords while preserving user privacy.

Hands-on workshop on developing applications with Google Vertex AI and Gemini 1.0 Pro.

Key points:

• Model deployment on Google Cloud Vertex AI
• Prompt engineering with Gemini 1.0 Pro
• Building production AI applications
• Integration with Google Cloud services

Takeaway: Vertex AI + Gemini provides a complete pipeline for deploying production AI applications on Google Cloud.

Hands-on session introducing function calling with Google Gemini API.

Key points:

• Connecting LLMs to external tools and APIs
• Building agentic applications with function calling
• Practical examples of Gemini function calling
• Best practices for prompt engineering with tools

Takeaway: Function calling bridges the gap between LLMs and real-world tools, enabling AI agents that can take actions beyond text generation.

Workshop at University of Missouri - Kansas City on giving web apps superpowers with Generative AI and MediaPipe.

Key points:

• On-device AI with Google MediaPipe for image processing
• Gesture recognition and hand tracking in the browser
• Real-time ML inference without server calls
• Combining Generative AI with web applications

Takeaway: MediaPipe brings production-quality ML to the browser, enabling real-time AI features without sending data to the cloud.

Hands-on workshop at University of Washington on LLM application components and design patterns.

Key points:

• RAG (Retrieval-Augmented Generation) architecture patterns
• Agent design patterns for tool use and planning
• Memory management for long conversations
• Building production LLM systems with proper evaluation

Takeaway: Production LLM applications require thoughtful architecture — RAG for knowledge, agents for action, and proper evaluation for reliability.

Talk at Google San Jose Meetup on building on-device generative AI in the browser.

Key points:

• Running image generation models in the browser with WebGPU
• TensorFlow.js for on-device ML inference
• Building a DALL-E-like image generator that runs locally
• Privacy benefits of on-device AI

Takeaway: WebGPU enables real AI computation in the browser — you can run generative models locally without sending data to any server.

Talk at Berkeley RDI presenting an empirical study evaluating the quality of smart contracts generated by LLMs.

Key points:

• Systematic evaluation of ChatGPT and Google Palm2 for Solidity code generation
• Analysis of validity, correctness, and security of AI-generated smart contracts
• Security bugs introduced by LLMs in generated code
• Impact of input quality and prompting strategies on output

Takeaway: While LLMs can generate functional smart contracts, they introduce security vulnerabilities that require careful review and testing before deployment.

Talk on DeepSpeech and the journey to achieving less than 10% word error rate in automatic speech recognition.

Key points:

• Mozilla DeepSpeech architecture and training
• Challenges in achieving low word error rates
• On-premise ASR deployment for privacy
• Real-world deployment at Fireflies.ai serving 10M+ users

Takeaway: Open-source ASR can achieve production-quality results while keeping speech data on-premise for privacy.

Talk on the GDE Stage at Google North America Connect 2022 on the state of Web Mixed Reality.

Key points:

• Evolution from WebVR to WebXR
• Current state of mixed reality on the web
• Google’s commitment to WebXR in Chrome
• GDE community contributions to web technologies

Takeaway: WebXR has replaced the separate WebVR and WebAR standards with a unified API, making mixed reality development simpler.

Talk at North Carolina DevFest 2022 on DeepSpeech architecture.

Key points:

• DeepSpeech neural network architecture explained
• Training pipelines for speech recognition models
• Optimizing for different languages and accents
• Live demo of on-device speech recognition

Takeaway: Open-source speech recognition has matured to the point where it can be deployed in production with accuracy rivaling commercial offerings.

Talk at NDC Sydney 2020 on the state of WebXR.

Key points:

• Unifying VR, AR, MR, XR — what do they all mean for developers?
• Why now? The convergence of browser support and hardware
• What you need to know to build mixed reality experiences
• Mozilla Mixed Reality team’s work on bridging platforms
• Building WebVR with A-Frame and WebAR without markers

Takeaway: The virtual world is cluttered with jargon, but WebXR simplifies it — one API for all immersive experiences, running in any browser.

FOSDEM 2020 JavaScript Devroom talk on building decentralized social VR with WebXR.

Key points:

• Building auditable, privacy-aware social VR in JavaScript
• Decentralized room hosting for censorship resistance
• IPFS integration for decentralized asset delivery
• Adapting to Mixed Reality capabilities of each device
• Cross-platform from phone to high-end VR headset

Takeaway: Decentralized social VR is technically feasible today using WebXR, IPFS, and open-source tools — enabling user-hosted, privacy-respecting virtual spaces.

Talk at FOSDEM 2020 on building auditable, privacy-aware social VR experiences using WebXR.

Key points:

• Building social VR spaces that run entirely in the browser
• Decentralized room hosting resistant to takedowns
• Privacy-aware design with auditable code
• Cross-platform support from mobile to high-end VR headsets
• IPFS integration for decentralized asset hosting

Takeaway: You can build fully functional, privacy-respecting social VR experiences using only JavaScript, WebXR, and decentralized protocols — no SDK lock-in required.

Talk at Selenium Conference 2020 on testing Web VR/AR Mixed Reality applications.

Key points:

• WebXR specification and the W3C Immersive Web Group
• Differences between testing regular web pages and Mixed Reality
• How to define “smooth” vs just “responsive” in VR
• Automation strategies for MR testing
• Performance metrics for immersive experiences

Takeaway: Testing Mixed Reality requires new approaches — traditional web testing tools need adaptation for spatial, immersive, and motion-based interactions.

Mozilla TechSpeaker Paris Meetup talk introducing jARvis, a voice-driven assistant.

Key points:

• Introduction to jARvis, a JavaScript-based voice assistant
• Voice interaction with web technologies
• Mozilla TechSpeaker program overview
• Live demo of voice-driven web interactions

Takeaway: Voice-driven web applications are buildable today with open web APIs — no proprietary assistants needed.

Lightning talk for Mozilla TechSpeaker Singapore Meetup on Voice Driven VR.

Key points:

• Voice interaction for VR experiences
• Web Speech API integration with WebVR
• Hands-free VR navigation and control
• Accessibility benefits of voice-driven VR

Takeaway: Voice-driven VR makes immersive experiences more accessible and natural — the Web Speech API brings voice to web-based VR.

Talk at DevFest Ranchi 2019 on DeepSpeech and the journey to less than 10% word error rate.

Key points:

• Mozilla DeepSpeech open-source ASR engine
• Training and fine-tuning speech models
• Achieving production-quality accuracy
• Deployment at Fireflies.ai for 10M+ users

Takeaway: Open-source speech recognition can match commercial accuracy while keeping data private and on-premise.

Talk at the Data Driven Innovation conference in Edinburgh on AI and data-driven web applications.

Key points:

• Using data-driven approaches for scalable web applications
• AI-powered analytics and visualization
• VR as a data visualization platform
• Open web standards for data-driven innovation

Takeaway: Data-driven innovation requires combining AI, web technologies, and open standards to create accessible and scalable solutions.

Talk at DevDay on WebVR and open web technologies.

Key points:

• Building cross-platform VR with JavaScript
• The WebXR API and browser support landscape
• A-Frame as a beginner-friendly VR framework
• Demos of browser-based VR experiences

Takeaway: The open web is a viable platform for VR development, offering cross-platform compatibility without vendor lock-in.

Tech talk at UT Dallas Welcome Week on WebVR and emerging web technologies.

Key points:

• Introduction to WebVR and the WebXR API
• Building VR experiences with A-Frame
• Career paths in web and VR development
• Open source contributions and community

Takeaway: Students can start building VR experiences today using web technologies they already know — no special equipment needed to get started.

Talk at O’Reilly OSCON 2018 on WebVR and Mozilla Mixed Reality. Video available on O’Reilly platform.

Key points:

• Bridging VR, AR, MR, and XR technologies with open web standards
• The WebXR API as a unified standard
• Building VR apps with A-Frame without vendor lock-in
• Mozilla’s role in the Immersive Web Working Group
• Live coding and demos

Takeaway: Open source VR through WebXR ensures immersive computing remains accessible to all developers, not just those with proprietary SDKs.

Tutorial at O’Reilly OSCON 2018 on building WebVR experiences.

Key points:

• Hands-on WebVR development with A-Frame
• Building interactive VR scenes from scratch
• Performance optimization for web-based VR
• Deploying VR experiences to any browser

Takeaway: Attendees left with a working VR experience built entirely with web technologies, ready to deploy.

Talk at RevolutionConf on WebVR and Mixed Reality.

Key points:

• The WebXR API and A-Frame framework
• Building immersive experiences for the open web
• Cross-platform VR without vendor lock-in
• Live coding demo

Takeaway: The web is the most accessible platform for VR development — no app stores, no SDK fees, just open standards.

Talk at OpenVis Conf on visualizing data in 3D VR worlds using A-Frame.

Key points:

• Data visualization in immersive 3D environments
• A-Frame for building data-driven VR scenes
• Interactive data exploration in VR
• Benefits of spatial data visualization over 2D charts

Takeaway: VR transforms data visualization — immersive 3D environments reveal patterns and relationships that flat charts cannot show.

Talk at Fullstack London on building browser-based VR experiences with WebVR and A-Frame.

Key points:

• WebVR API and browser support
• A-Frame framework for declarative VR development
• Performance optimization for web-based VR
• Cross-platform deployment strategies

Takeaway: Browser-based VR is production-ready and accessible to any web developer willing to learn the WebXR API.

Talk at Immersive Technology Conference on the state of WebVR, A-Frame, and Mozilla’s work on Web Mixed Reality.

Key points:

• Journey through VR history to the current Mixed Reality landscape
• WebVR API and A-Frame framework for browser-based VR
• Building VR and AR experiences with JavaScript
• Use cases, real-life applications, and case studies
• Live coding and demos

Takeaway: The past, present, and future of mixed reality is on the web — A-Frame and WebVR make it accessible to any web developer.

Talk at DevWeek NYC 2018 on the state of WebXR and building web mixed reality applications.

Key points:

• Current state of WebVR, WebAR, and the unified WebXR API
• What developers need to know to build mixed reality apps
• Browser support and device compatibility
• Mozilla Mixed Reality team’s contributions

Takeaway: WebXR is the unified standard for mixed reality on the web — developers should start building with it now.

Talk at DeveloperWeek Austin 2017 on mixing real and virtual in WebAR.

Key points:

• Augmented and mixed reality for everyone using web standards
• WebAR without markers using the WebXR API
• Mixing real-world camera input with virtual objects
• Cross-platform AR without SDK lock-in

Takeaway: WebAR brings augmented reality to billions of devices through the browser, no app installation required.

Talk at Mobile Day 2017 on VR for Humans using WebVR and A-Frame.

Key points:

• Building VR experiences without expensive ecosystems or new technology stacks
• Creating immersive VR with the same toolchain web developers already know
• A-Frame for declarative VR development
• Cross-platform VR from mobile to high-end headsets

Takeaway: VR is for humans — not just developers with expensive hardware. WebVR and A-Frame make it accessible to anyone who can build a web page.

Talk at Mexico City Mobile Day on VR for Humans using WebVR and A-Frame.

Key points:

• Building VR experiences for mobile devices
• WebVR and A-Frame for cross-platform VR
• Creating immersive worlds with JavaScript
• No expensive hardware or SDKs needed

Takeaway: VR is for everyone — with WebVR and A-Frame, any mobile developer can create immersive experiences that run in a browser.

Attended and showcased WebVR at Google Chrome Dev Summit.

Key points:

• WebVR demonstrations and discussions with Chrome team
• Browser-level VR support and performance
• Future of WebXR in Chrome
• Networking with web platform engineers

Takeaway: Chrome’s commitment to WebXR signals that browser-based VR is a first-class platform, not an experiment.

Talk at MIT Media Lab on turning sensors into signals and freeing IoT from walled gardens.

Key points:

• Humanizing IoT with old smartphones and web technologies
• Using JavaScript to process sensor data on the edge
• Privacy-aware IoT design avoiding cloud dependency
• Open web standards as an alternative to proprietary IoT platforms
• Freeing IoT from walled gardens with open protocols

Takeaway: You don’t need expensive IoT platforms — old smartphones and JavaScript can create privacy-aware, open IoT systems that keep data local.

Talk at LinuxCon on hardening IoT endpoints with a preventive toolkit. Slides available.

Key points:

• Security risks of default embedded Linux configurations
• A cookbook toolkit for hardening Raspberry Pi and similar devices
• Audit and behavioral analysis for continuous monitoring
• Code examples and live demonstrations
• Preventing device takeover and botnet recruitment

Takeaway: IoT security is not optional — the toolkit provides a systematic, code-level approach to hardening embedded Linux devices against common attacks.

Talk at LinuxCon ContainerCon on hardening IoT endpoints with a preventive toolkit.

Key points:

• Security risks of default Raspberry Pi configurations
• A cookbook toolkit for hardening embedded Linux devices
• Audit and behavioral analysis techniques
• Preventing botnet recruitment (e.g., Mirai-style attacks)

Takeaway: Securing IoT devices is critical — a single unpatched Raspberry Pi can become part of a botnet. The toolkit provides a systematic approach to hardening.

Presentation at Singapore Leadership Summit on Mozilla Connected Devices and TechSpeaker program.

Key points:

• Mozilla’s Connected Devices strategy and vision
• Privacy-first IoT and smart home technology
• TechSpeaker program for community evangelism
• Open standards for connected devices in Asia

Takeaway: Mozilla’s approach to connected devices demonstrated that IoT can be both useful and privacy-respecting.

Talk at Kansas City Developer Conference 2017 on building WebVR with A-Frame.

Key points:

• WebXR API and A-Frame framework introduction
• Building VR scenes with declarative HTML
• Cross-platform VR from mobile to desktop
• Live coding demo

Takeaway: A-Frame makes VR development as approachable as web development — if you know HTML, you can build VR.

Talk at Open Source Bridge 2017 on open web and VR technologies.

Key points:

• WebVR and A-Frame for open VR development
• Building immersive experiences with web standards
• No vendor lock-in for VR content
• Community-driven VR framework

Takeaway: Open source VR through A-Frame ensures VR content creation remains accessible and不被 locked into any platform.

Talk on optimizing WebVR performance, from the MS thesis at Rice University.

Key points:

• WebGL rendering optimization for VR workloads
• Three.js performance tuning techniques
• Browser-level JavaScript parallelization for VR
• Offloading computations to GPU with WebAssembly
• Object loading and texture rendering optimization

Takeaway: WebVR can achieve near-native performance with careful optimization of the rendering pipeline, GPU offloading, and browser-level parallelism.

Mozilla Connected Devices Team presentation at MozLondon.

Key points:

• Mozilla’s vision for connected devices and IoT
• Privacy-first approach to smart home technology
• Open standards for IoT communication
• Community-driven IoT projects

Takeaway: Mozilla’s connected devices initiative demonstrated that IoT can be privacy-respecting and open, unlike many commercial offerings.

Presentation at ACM SIGGRAPH Web3D on optimizing WebVR performance.

Key points:

• WebGL rendering pipelines for VR applications
• Three.js optimization techniques for complex scenes
• Browser-level parallelism for VR workloads
• Performance benchmarks across different VR headsets

Takeaway: WebVR can achieve native-like performance with careful optimization of the rendering pipeline and leveraging browser-level GPU acceleration.

Talk at Linux Foundation Open IoT & ELC on hardening Raspberry Pi security.

Key points:

• Security risks of default Raspberry Pi configurations
• A toolkit and cookbook for hardening Pi devices
• Techniques for preventing botnet recruitment
• Continuous audit and behavioral analysis
• Code examples and live demonstrations

Takeaway: Protecting your IoT device protects everyone — an unsecured Raspberry Pi can become part of a botnet like the Mirai DDoS attack. The IronClad toolkit makes hardening systematic.

Talk at Open Networking Summit on the state of WebVR and A-Frame.

Key points:

• Evolution of WebVR from early experiments to the WebXR standard
• A-Frame as a declarative framework for VR development
• Yesterday: early WebVR prototypes
• Today: production-ready WebXR in all major browsers
• Tomorrow: mixed reality and beyond

Takeaway: WebVR has come a long way from experimental prototypes to a production-ready standard — the future is mixed reality on the open web.

Talk at Silicon Valley Virtual Reality meetup on VR for Humans with WebVR.

Key points:

• Building VR worlds with WebVR and A-Frame
• Creating immersive experiences for any browser
• Cross-platform VR from mobile to high-end headsets
• Open web as an alternative to proprietary VR platforms

Takeaway: The open web is the most democratic platform for VR — anyone with a browser can experience and create VR content.

Talk at LibrePlanet 2017 on open source, VR, and the Mozilla Mixed Reality team.

Key points:

• Making VR accessible through open web standards
• WebVR and A-Frame as open alternatives to proprietary VR SDKs
• Mozilla’s role in keeping the web open for VR
• Community contributions to WebVR

Takeaway: Open source VR through WebVR and A-Frame ensures that immersive computing remains accessible and不被 vendor-locked.

Talk at Securi-Tay 2017 on security, covering static analysis and systems security.

Key points:

• Static analysis for code security
• Security challenges in emerging web technologies
• IoT device security hardening
• Model checking and proof-carrying code

Takeaway: Security must be built into development from the start — static analysis and formal methods can catch vulnerabilities before deployment.

Talk at Linux Foundation OpenIoT Summit on IoT and connected devices.

Key points:

• Using JavaScript and web technologies for IoT applications
• Privacy-aware IoT design principles
• Connecting sensors to the web platform
• Open standards for IoT communication

Takeaway: The web platform provides a privacy-respecting, vendor-neutral foundation for IoT development using open standards.

Talk at DeveloperWeek 2017 on building WebVR applications using A-Frame.

Key points:

• Introduction to the WebVR API and A-Frame
• Building VR scenes with HTML-like syntax
• Cross-platform VR deployment from mobile to desktop
• Live coding demo of a VR experience

Takeaway: Web developers can create VR experiences using skills they already have — HTML, CSS, and JavaScript — with A-Frame.

Workshop at Linux Foundation OpenIoT Summit on building virtual reality worlds using WebVR and A-Frame.

Key points:

• Creating immersive VR scenes with A-Frame entities
• JavaScript-based VR development without game engines
• Running VR experiences on any device with a browser
• Connecting IoT data streams to VR visualizations

Takeaway: WebVR and A-Frame make VR development accessible to web developers, no specialized game engine knowledge needed.

A Mozfest session exploring the privacy implications of connected IoT devices.

Key points:

• How smart devices collect behavioral and environmental data
• Privacy concerns around data retention and observation
• Ethical implications of always-on connected devices
• User awareness and control over their data

Takeaway: Users should understand what their connected devices know about them and demand transparent, privacy-respecting IoT designs.

Workshop at Mozfest on painting with VR using A-Painter.

Key points:

• Hands-on VR painting with A-Painter (Mozilla’s Tilt Brush alternative)
• Creating 3D art in the browser with WebVR
• Sharing VR paintings online
• No expensive software needed — just a browser and VR headset

Takeaway: VR painting is accessible to everyone through the browser — A-Painter brings creative VR tools to the open web.

Talk at GraphicalWeb 2016 on web graphics and VR.

Key points:

• WebGL rendering optimization techniques
• Three.js performance for complex scenes
• The emerging WebVR standard
• Browser GPU acceleration for graphics

Takeaway: The web is becoming a serious platform for high-performance graphics and VR, powered by WebGL and emerging WebVR standards.

Game demo and VR booth at ViewSource Berlin.

Key points:

• Hands-on VR demos built with WebVR and A-Frame
• Showcasing browser-based VR games
• Community engagement with Mozilla technologies
• Networking with European web developers

Takeaway: Browser-based VR games are fun and functional — ViewSource Berlin showcased the creative potential of WebVR.

Speaker at Mozilla TechSpeaker Berlin Meetup.

Key points:

• Mozilla TechSpeaker program for community evangelism
• Web technologies and emerging standards
• Community building through tech talks
• Open web advocacy

Takeaway: The Mozilla TechSpeaker program empowered community members to spread open web technologies globally.

Talk at BrazilJS on the state of WebVR and building immersive experiences with A-Frame.

Key points:

• Introduction to the WebVR API and A-Frame framework
• Building cross-platform VR experiences with JavaScript
• Demo of social VR experiences running in the browser
• The future of WebXR and mixed reality on the web

Takeaway: WebVR democratizes VR development — anyone with JavaScript skills can build immersive experiences that run on any device with a browser.

Talk at Open Source Bridge 2016 on open web technologies.

Key points:

• WebVR, A-Frame, and the Mozilla Mixed Reality ecosystem
• Open standards for immersive web experiences
• Community contributions to the WebVR specification
• Demos of browser-based VR

Takeaway: The open source community is driving the future of immersive web experiences through WebVR and A-Frame.

Talk at Open Source Hong Kong on open web and VR.

Key points:

• The WebVR standard and browser support
• A-Frame framework for VR development
• Building immersive experiences with JavaScript
• Open web vs proprietary VR platforms

Takeaway: The open web provides a universal platform for VR that works in Hong Kong as well as anywhere else — no app stores, no gatekeepers.

Talk on building a self-learning word prediction module for multilingual input.

Key points:

• Language input challenges on mobile devices
• Predictive text that learns from user behavior
• Supporting localized languages (Hindi, Bengali)
• Implementation as an open web app for Firefox OS
• Transliteration and IME challenges

Takeaway: Self-learning prediction adapts to each user’s language and typing style, outperforming static dictionaries for multilingual input.

Talk at JSFoo 2015 on building a self-learning word prediction module for Firefox OS.

Key points:

• Predictive text input for mobile devices
• Self-learning from user typing patterns
• Handling multilingual input (Hindi, Bengali)
• Implementation in JavaScript for Firefox OS
• Challenges with transliteration systems

Takeaway: Self-learning word prediction can significantly improve typing experience for localized languages, adapting to each user’s patterns over time.

Talk at Open Source Bridge 2015 on building a self-learning word prediction and auto-correct module for Firefox OS.

Key points:

• Predictive text for mobile devices using JavaScript
• Self-learning from user typing patterns
• Handling multilingual input (Hindi, Bengali, IME languages)
• Implementation in Firefox OS as an open web app
• Transliteration challenges and solutions

Takeaway: Self-learning word prediction works for any language — the open web enables predictive text that adapts to users, not the other way around.

Talk at Open Source Bridge 2015 on building a self-learning word prediction module.

Key points:

• Self-learning word prediction for multilingual input
• Implementation for Firefox OS in JavaScript
• Adapting to user typing patterns over time
• Challenges with localized languages (Hindi, Bengali)

Takeaway: Open source predictive text can adapt to users and languages that commercial offerings often ignore.

Talk at WikiConference USA 2014 on open knowledge and web technologies.

Key points:

• Contributing to Wikipedia and open knowledge platforms
• Web technologies for collaborative content creation
• Open source tools for the knowledge community
• Bridging technical and content communities

Takeaway: Open knowledge platforms like Wikipedia rely on both content contributors and technical contributors — both are essential.

Talk at Confluence 2012 on QA Robotics and innovation in automated testing.

Key points:

• Automated mobile hardware and software testing with robotics
• OpenCV-based vision systems for QA
• Innovation in testing frameworks at Cognizant
• Live demo of robotic QA testing

Takeaway: Robotics combined with computer vision can automate hardware QA testing that would be impossible manually.