TNaaS - Tech Names as a Service
summary
A web service generates Pokemon‑like names and actually runs at https://dyccxmwpz9.execute-api.eu-west-1.amazonaws.com/pokemon/poke-names/v2 with the domain tnaas.com purchased; models compared include simple Markov chains, probabilistic graphical models implemented with pomegranate and proposals for RNN/LSTM and bidirectional completion, but deep learning is avoided due to a small corpus (~700 Pokemon names). Deployment uses AWS Lambda + API Gateway, yields ~200 ms cold-to-warm responses, is very cheap to operate, but needs caching, throttling and contingency planning for cloud outages.
AI-written summary · about
chapters (16)
- 0:00Introduction to Tech Names as a Service
- 2:31Project inspiration
- 4:37Live demo
- 6:43Name generation math
- 9:12Other naming use cases
- 10:37Beyond Markov chains
- 13:22Why deep learning fails
- 15:07Improving user experience with input
- 17:13Graphical models explained
- 18:40AWS Lambda deployment
- 21:25Lambda setup guide
- 24:54Cost comparison
- 27:02Security and scalability
- 28:24Lambda performance tips
- 30:10CLI deployment automation
- 31:55Conclusion
related talks
33mAdvanced Natural Language Processing with Free Low-Code and...
17mNatural Language Processing in Automation
15mAI as a Service: Quick and Scalable Machine Learning POCs with No-Code/Low-Code Tools
36mAccessible Neural Search and Cloud-Nativeness for ML Engineers
01h 18mBuild Data Apps by Deploying ML Models as API Services
39mFrom data science to scalable NLU and vision cloud service.
34mPatterns of Application Development Using AI
42mPrompt. Build. Repeat. – The Vibe Coding Mindset