Alumnx AI Build Day: TokenLens

​Alumnx AI Build Day: Token Observability

​🗓️ 30th May | ⏰ 11 AM – 12 PM 📍 Hybrid — Hyderabad + Online

​Ever wondered what your AI inference actually costs?

​When you upload a PDF into ChatGPT or Claude and ask a question — do you know how many tokens are being consumed? Do you know when it makes more sense to self-host an open source model like Gemma 4 versus calling a hosted API like Gemini Flash?

​Most people building AI applications today don't have answers to these questions. That's an expensive blind spot.

​What we are building

​Participants from Alumnx AI Labs' Full Stack AI Engineers cohort and AI for Senior Leaders (SLP) program are building a Token Observability web application — live, in 48 hours.

​The application will let you:

• ​Upload a PDF and ask questions against it

• ​See exactly how many tokens are consumed and what it costs — in ₹ and $

• ​Compare full-context ingestion vs RAG-based retrieval

• ​Understand when a self-hosted model beats a hosted API — and when it doesn't

​What happens on 30th May

​We demo the live, deployed application — built entirely by our cohort in 48 hours. You will see real token counts, real cost breakdowns, and a side-by-side comparison of Gemma 4 on an NVIDIA A10G GPU versus Gemini Flash API.

​Who should attend

• ​Developers building AI applications who want to understand and control their inference costs

• ​Senior leaders evaluating AI investments and ROI

• ​AI enthusiasts curious about how LLMs are priced and deployed

• ​Anyone who wants to see what 48 hours of focused AI building looks like

​Want to build with us?

​We are open to contributions from AI enthusiasts who want to code alongside our team during the 48-hour build sprint. Limited spots available.

​Organised by Alumnx AI Labs | Hyderabad Building hands-on AI practitioners — one cohort at a time.