Building SLM: Small Models, Big Impact
The AI landscape is obsessed with "Bigger." But the real revolution is happening in "Smaller."
Join us for the next edition of AI Weekends hosted by Kovan Labs.
While the industry debates massive, expensive language models, Small Language Models (SLMs) are quietly becoming the heroes of real-world deployment. They offer low latency, high privacy, cost-effective scalability, and edge capabilities.
In this deep-dive masterclass, we are exploring how compact, optimized models (like Phi-3, Llama-3 8B, and Mistral) are creating oversized impacts across industries.
The Workshop Agenda
Part 1: The AI Landscape & The SLM Shift
Beyond the Hype: Why "bigger" isn't always better. Understanding the interplay between LLMs and Autonomous Agents.
Deep Dive into SLMs: What defines a "small" model? Exploring the architecture and parameter counts of today's best compact models.
The Efficiency Frontier: How SLMs work under the hood to deliver low latency, reduced compute costs, and local/edge deployment.
Part 2: Strategy & Selection
The "Best Fit" Framework: How to benchmark accuracy vs. speed, and choose the right model for reasoning, coding, or writing.
Building a Product-First AI Strategy: Leveraging SLMs to build intuitive, snappy user experiences without relying on expensive, slow cloud APIs.
Part 3: Data & Fine-Tuning
Curating Your Foundation: Where to find high-quality datasets (Hugging Face, synthetic data) and how to clean them.
Tailoring the Intelligence: Fine-Tuning 101. An introduction to techniques like LoRA and QLoRA to adapt a general model to your industry jargon.
☕ 15-Min Networking & Coffee Break
Part 4: Hands-On & Implementation
The Fine-Tuning Lab: A practical walkthrough of taking a base SLM and optimizing it for a specific task.
Training Infrastructure: Navigating local GPUs, Google Colab, and cloud clusters.
Deployment (The Final Mile): How to host your model, create an inference API, and push it to a production environment.
🎙 About the Speaker
Sanju Sivalingam Founder @ dun | Founder & CEO @ ThisUX | Co-founder @ Unitedby.AI |
Sanju operates at the rare intersection of engineering and product design—a discipline he calls User Experience Engineering (UXE). As the founder of the global product studio ThisUX and the creator of the recently launched dun social, Sanju has deep, hands-on experience bridging the gap between raw technical capability and user-centric design.
He will be sharing real-world perspectives on how building with SLMs allows you to create highly responsive, cost-effective, and scalable AI products like dun. He is also the creator of the acclaimed open-source project droidclaw.ai.
🧠 Who Should Attend?
Developers & Engineers wanting to learn how to run and fine-tune models locally or on edge devices.
Product Managers & Founders looking to build cost-effective, high-margin AI features without burning cash on API calls.
UX/UI Designers curious about how model latency and architecture impact the final user experience.
⚡ Prerequisites
Bring your Laptop & Charger. (Required for the Part 4 hands-on lab).
A basic understanding of AI concepts (You don't need to be an ML researcher, but you should know what a prompt or an API is).
📍 Venue: Kovan Labs 2nd Floor, IndiQube Echo, Coimbatore
🎟 Admission: Free. Space is limited. Priority is given to builders, developers, and active learners.