WTA AT “APPLYING LLMs IN PRACTICE”: TURNING FOUNDATIONS INTO FUNCTIONAL AI TOOLS

Harvard SEAS – Allston Campus, February 21–23, 2025

In February 2025, Western Technology Alliance (WTA) had the privilege of sending both our CEO, Jeremy Kahn, and Chief Technology Officer, Jerry Bradenbaugh, to “Applying LLMs in Practice: Core Concepts and Functional AI Solutions for Real-World Implementation,” a three-day workshop hosted by Harvard SEAS at its Allston campus. The event brought together technology leaders, researchers, and industry professionals to explore how large language models (LLMs) can move from theory into impactful, scalable solutions.

Led by Harvard’s Pavlos Protopapas and a cohort of experts, the program focused on demystifying the architecture, training, and deployment of LLMs while diving deep into hands-on exercises and applied case studies.

From the Lab to the Field

Throughout the immersive sessions, WTA leadership engaged in practical labs, peer discussions, and expert-led modules that covered:

• Transformer fundamentals: Including tokenization, embeddings, and attention mechanisms

• Advanced workflows: Such as retrieval-augmented generation (RAG), fine-tuning techniques, and model alignment

• Enterprise use cases: Including AI-driven customer support, document summarization, and dynamic content creation

A core takeaway was the importance of aligning business objectives with technical solutions, ensuring each implementation is targeted, traceable, and optimized for operational efficiency. The program also emphasized the balance between using general-purpose foundation models and customizing outputs through strategic fine-tuning.

WTA’s Strategic Takeaways

The workshop’s real-world focus directly informed WTA’s AI and engineering strategy. Some of the most immediate implications include:

• Product enhancements: We're actively integrating LLMs into client-facing applications, particularly for automated reporting, intelligent form generation, and contextualized search.

• Technical infrastructure: We're evaluating hosting options for in-house and third-party model orchestration, particularly for RAG workflows.

• Workforce upskilling: WTA is expanding internal education efforts for staff across engineering, QA, and product management teams.

• Responsible AI adoption: The event reaffirmed our commitment to data transparency, bias mitigation, and ethical AI deployment in all phases of the development lifecycle.

Looking Ahead to Scalable Impact

WTA is launching several pilot initiatives in 2025. These include:

• Interactive client-facing tools for real-time question answering

• Streamlined workflows for compliance report generation using LLMs

• Deployment of custom agents for intelligent task routing and triage

Our focus remains on measurable results: reduced turnaround times, improved document quality, and enhanced end-user experience.

A Commitment to Continued Advancement

WTA’s participation in “Applying LLMs in Practice” is part of a broader investment in staying ahead of the AI curve. With real-time experimentation, community-driven innovation, and ethical oversight at the center of our roadmap, we are building solutions that are not only cutting-edge but also responsible and user-focused.

We extend our thanks to the organizers, instructors, and fellow participants at Harvard SEAS for an inspiring and highly practical experience.