Shubhangi Goyal

I presented an in person talk on AI agents at AgentCon in Sofia, Bulgaria. The talk explored how AI agents are designed to use context effectively to execute complex tasks and deliver real business value. It covered key aspects such as context engineering, memory management, tools, and task orchestration to enable adaptive behaviour. I also discussed how agent architectures and workflows operate, the role of prompt engineering within the broader system, and common pitfalls in deploying AI agents, supported by real-world use cases. I also discussed practical frameworks and real-world design patterns to build AI applications with high contextual intelligence. Attendees gained practical insights into designing robust agent systems, understanding how components work together, and avoiding common implementation challenges when building scalable, context-aware AI solutions.

I presented an in person session at DevFest Bletchley Park on LLM evaluation and its use cases. In this session, I discussed how prompt engineering can be leveraged to evaluate the performance, reliability, and limitations of Large Language Models (LLMs). By systematically varying prompts, I showed examples of model behavior across various tasks, including reasoning, summarization, and code generation. I also discussed benchmarking methods, prompt sensitivity, and how to design effective evaluation frameworks. Key takeaways for the attendees were: - Understand the role of prompt design in LLM evaluation. - Learn techniques to test reasoning, bias, and robustness via prompts. - Explore limitations of prompt-based metrics and how to mitigate them. - Gain practical examples of evaluation setups across NLP tasks. This session equipped attendees with knowledge of LLM models evaluation methods, and techniques along with real world examples.

I presented a talk on AI agents at GenAI London 2025 Conference in London. The talk explored how AI agents are designed to use context effectively to execute complex tasks and deliver real business value. It covered key aspects such as context engineering, memory management, tools, and task orchestration to enable adaptive behaviour. I also discussed how agent architectures and workflows operate, the role of prompt engineering within the broader system, and common pitfalls in deploying AI agents, supported by real-world use cases. Attendees gained practical insights into designing robust agent systems, understanding how components work together, and avoiding common implementation challenges when building scalable, context-aware AI solutions. The session helped them understand how to architect scalable, trustworthy AI agents that perform effectively in real business environments.

I presented a talk at Civo Navigate Conference in London. In this talk I discussed about transformative role of Agentic AI in reshaping decision-making processes and about agentic architecture, the process and ways to drive efficiency. This session explored agentic architectures, how agents operate across workflows, and the ways in which they drive efficiency, speed and innovation in modern systems. I also highlighted considerations such as bias, accountability and transparency and discussed strategies for balancing AI with responsible decision making. Attendees gained a clear understanding of how to design and implement AI systems in real-world environments. It enabled participants to think beyond traditional AI models and adopt a more structured, responsible approach to building intelligent, decision-driven systems. It also helped them to understand practical frameworks and design patterns to build applications with contextual intelligence.

I actively mentor aspiring professionals in data and generative AI, helping them navigate both technical challenges and career growth. Through platforms like Femme Palette, I guide mentees on topics ranging from data analysis, machine learning, and model evaluation to understanding generative AI tools like large language models. I focus on simplifying complex concepts, sharing real-world applications, and providing feedback on projects, portfolios, and interview preparation. My approach is tailored—whether someone is switching into data, advancing their skills, or exploring the potential of AI, I meet them where they are. Beyond technical mentoring, I also help build confidence, clarify career paths, and connect mentees with resources and communities. I believe in fostering curiosity, ethical thinking, and a hands-on mindset. By mentoring in both data and generative AI, I aim to shape capable, responsible professionals who can innovate with purpose.