Aprajita Verma
Frontend Architect
Delhi, India
Actions
Aprajita Verma is a Women Techmakers Ambassador and Frontend Architect with 17 years building web and mobile products that hold up at scale, at MYCOM. She builds AI-first apps where accessibility and user experience are requirements, not polish, and now works mostly in Agentic AI and Agentic UX, designing interfaces where AI is a collaborator, not just a feature. A renowned voice on the tech speaking circuit, she has spoken at numerous national and international conferences and mentors engineering teams at meetups across India.
Speaker at Devfest Chandigarh 2025, Devfest Noida 2025, Devfest Patna 2025, City JS India 2026 and React India 2025 remote.
ngIndia 2026 and City JS London 2026 https://www.linkedin.com/in/aprajita-verma-19522814a
Area of Expertise
Topics
Agent-Shaped Frontend: The Missing Links
What changes in frontend architecture when the backend is no longer just an API, but an autonomous agent? This talk dives into the missing primitives for agent-native UIs: streaming agent state, tool-call lifecycles, event-driven state management, human-in-the-loop checkpoints, optimistic execution, partial results, and failure recovery. We’ll connect agent orchestration patterns with React frontend architecture and explore how protocols like MCP reshape the boundary between UI, tools, and agents.
404: Backend Not Found, but AI survived
I wanted to find out how much of an AI app can run in the browser with no backend, so I built a live demo for every layer I could: Chrome's built-in Gemini Nano through the Prompt API, TensorFlow.js and MediaPipe for vision, WebGPU and WebAssembly for compute, and Transformers.js, WebLLM and ONNX Runtime Web for bigger models, with OPFS holding the weights. A lot of it worked. A lot of it broke in ways no tutorial warned me about.
This talk is the breakage log, one failure at a time. WebLLM's model weights redirect from Hugging Face to a CDN and fail the browser's cross-origin check. The moment the page sets the COOP and COEP headers that SharedArrayBuffer needs, every asset from another origin gets blocked (Pyodide from jsDelivr, MediaPipe models from Google Cloud Storage, MobileNet from tfhub.dev), and without those headers, multithreaded WebAssembly quietly falls back to one thread. A TensorFlow.js model that fetched fine and then died with "Cannot read properties of undefined (reading 'producer')" because it was the wrong model format. Gemini Nano, which is unavailable without the right Chrome flags and a one-time model download, and whose API name has changed along the way. And the UI problem nobody mentions: a model that downloads silently looks exactly like a broken app.
You will see each failure live, with the cause and the fix, and with the WiFi off at the end, a demo or two might not survive the room. You will leave with a field guide to what is actually safe to ship in the browser today, a decision framework for on-device, server or hybrid AI, and the fixes for the traps you would otherwise hit on day one.
The Cheapest Token (Is the One You Never Send)
Generative AI is part of daily work, but few of us know what it costs the planet: electricity, CO2, the water used for cooling and the hardware behind every request. This talk is a practical guide to frugal AI for frontend and product teams. We start with what the footprint actually is, why published estimates vary so widely, and what you can realistically measure yourself: tokens sent, calls made and calls avoided. Then the actions. Decide which requests can be answered on the device instead, using Chrome's built-in Gemini Nano and small models in the browser. Route the rest through a hybrid setup so a server model is the exception, not the default. Trim what you do send with tighter prompts, caching and response limits. We also cover the rebound effect, where cheaper AI simply leads to more AI use, and why frugality needs a budget, not just good intentions. Live demos show features running with zero API calls. You will leave with a decision checklist for on-device, hybrid or server, and a way to make the case to the people who pay the bill.
AI + UX + Angular: Crafting Dashboards That Think Before They Render
Over the past year, I have been reimagining Angular dashboards that do not just visualize data, they think with you. This session offers a behind-the-scenes look at that journey.
We will explore how Google’s Gemini AI and Firebase transform static dashboards into intelligent, adaptive experiences.
We will see how Gemini generates natural-language insights and smart layout suggestions, while Firebase delivers real-time updates directly into Angular components.
We will then apply classic UX laws, Fitts’s for effortless interactions, Hick’s for quicker decisions, and the Law of Proximity for intuitive grouping, to make interfaces feel naturally responsive and human-centric.
Through live demos and real project patterns, the audience will learn how Google’s ecosystem can help build dashboards that combine AI reasoning, real-time data, and design psychology into one seamless experience.
AI-Proofing Your UI: Designing Frontends That Work for Humans and Machines
Users aren’t alone in our app anymore - AI assistants, smart browsers, and automated agents are also “reading” our UI.
But most frontends are still designed only for humans. In this talk, we’ll explore how to AI-proof our UI so it works beautifully for both humans and machines.
We’ll look at how structure, semantics, accessibility, and performance signals (like Core Web Vitals) affect how Chrome, search, and AI systems interpret pages.
We’ll learn practical patterns to make our Angular UI faster, more accessible, and more machine-understandable - without needing to be an AI engineer.
There has to be important discussion regarding usability laws which will still dominate the UI in the world of AI. We will discuss them as well through live demo.
Devfest Gandhinagar 2026 Sessionize Event
DevFest Jalandhar 2026 Sessionize Event
GDG Cloud Chandigarh - DevFest & Code for Communities 2026 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