Fireflies.ai Transcript & Audio Export + Organization for Future AI Project

# AI / RAG Developer to Analyze Tutoring Transcripts & Explore AI Applications ## About the Project We are an established education company that provides exam preparation and private tutoring for occupational therapy students preparing for the NBCOT® exam. Over time, we have accumulated transcripts from real tutoring sessions between our tutors and students. We currently have approximately **53 tutoring transcripts**, and we expect this library to continue growing. We believe these conversations may contain valuable proprietary knowledge about: * Questions students commonly ask * Concepts students struggle to understand * Common misconceptions * Test-taking mistakes * How experienced tutors explain difficult concepts * Teaching strategies that help students understand and apply information * Patterns among students who are struggling to pass their exam We want to determine whether this information can be leveraged using AI. **We are intentionally starting small.** We do not want to spend significant time or money building a large AI system until we have demonstrated that our tutoring data can provide meaningful value. We are looking for an experienced AI/LLM developer who can help us explore the possibilities, analyze our existing data, and build a small proof of concept. ## Initial Project We believe our approximately 53 tutoring transcripts are currently stored in Google Drive. The first objective is to analyze these transcripts using AI and determine what useful information and patterns can be extracted from them. We would like to explore questions such as: * What questions do students ask most frequently? * What concepts or topics cause the most difficulty? * What misconceptions occur repeatedly? * What test-taking mistakes are students making? * What strategies are our tutors repeatedly teaching? * What explanations or approaches do our tutors use to help students understand difficult concepts? * What patterns exist among students struggling with the NBCOT exam? * What information from these conversations could help us improve our existing educational program? * What new educational content should we consider creating? * What AI applications could potentially be developed from this data? We are also very interested in **AI use cases that we have not considered**. We want someone who can look at the data and help us determine where AI could create the greatest value for our company and students. ## AI Proof of Concept If the transcript analysis indicates that the data is valuable, we would like to create a **simple proof of concept** that allows us to interact with the knowledge contained in these tutoring sessions. For example, we might want to ask: "What mistakes do students commonly make when answering neurological questions?" "Which concepts are students struggling with most frequently?" "How do our tutors teach students to choose between two similar answer choices?" "What questions do repeat test takers frequently ask?" The system should be able to retrieve relevant information from our proprietary transcripts and provide useful responses grounded in our own data. A RAG-based knowledge system may be appropriate, but **we are open to recommendations from the developer**. This does not need to be a polished application or student-facing product. At this stage, we simply want to determine: **Does this work, and is it valuable enough to justify further investment?** ## Potential Future Projects If the initial project is successful, there could be significant additional work. Future projects could potentially include: * An internal AI knowledge assistant * Student-facing AI tutoring * AI tools that assist our human tutors * Automated analysis of future tutoring sessions * Identification of individual student weaknesses * Tutor training and quality-assurance tools * Personalized study recommendations * Integration of our tutoring knowledge with our proprietary educational content * Integration of AI into our existing educational platform * Analysis of thousands of student questions and interactions * Other AI applications identified through this project We are looking for someone who could potentially become a **long-term AI development resource for our company** if there is a strong fit. However, we want to prove the value through a small initial project before making a larger investment. ## What We Are Looking For Strong experience with: * Generative AI / LLM applications * OpenAI API and/or other leading LLM APIs * Retrieval-Augmented Generation (RAG) * Python * Embeddings and vector search * AI knowledge bases * Unstructured text/transcript analysis * API integrations Experience with Google Drive integrations and Fireflies.ai is helpful but not required. Experience in occupational therapy or NBCOT exam preparation is **not required**. We can provide the necessary subject-matter context. ## What Matters Most to Us We are not looking for someone who simply knows how to call an LLM API. We are looking for someone who can: * Understand a business problem * Recommend practical AI solutions * Work efficiently * Avoid unnecessary complexity * Build inexpensive prototypes before recommending larger investments * Explain technical concepts in plain English * Identify AI opportunities we may not have considered * Potentially grow with us as our AI needs expand We value **practical results over complicated architecture**. ## Initial Time Commitment We would like to begin with approximately **5–10 hours of work**. The initial goal is to review and analyze the transcript dataset, recommend the strongest potential AI use cases, and begin developing a simple proof of concept if appropriate. Please do not exceed the agreed initial number of hours without our approval. If the initial work is promising, we would be interested in expanding the project. ## Data Privacy These transcripts contain private educational conversations. Anyone working with this data must agree to keep it confidential and use it only for this project. Our data should not be: * Shared with unauthorized third parties * Used to train unrelated models * Retained for other purposes * Uploaded to unapproved services We would also appreciate recommendations regarding appropriate de-identification and data-handling practices as we expand this project. ## When Applying Please answer the following questions: 1. How would you approach analyzing approximately 53 tutoring transcripts to determine whether they contain useful knowledge for an AI application? 2. If you were asked to build a simple AI system that could answer questions using information from these transcripts, what approach would you recommend? 3. Please describe one RAG, AI knowledge-base, transcript-analysis, or similar LLM project you have personally built. 4. What AI tools/models would you consider for this project, and why? 5. How would you protect confidential student information when analyzing these transcripts? 6. What would you realistically expect to accomplish within the first 5–10 hours? 7. After reading this description, are there any potential AI use cases for these tutoring conversations that you think we may be overlooking? Please answer these questions in **plain English**. We are more interested in how you think and solve problems than in a long list of AI technologies. ## Compensation **Hourly: $30–$60/hour** We are willing to pay competitively for someone with strong AI/LLM experience who can work efficiently and help us identify the right opportunities. The initial engagement will be small, but **there is potential for significant ongoing work for the right person if the proof of concept demonstrates value.**