Hi, I'm Uday Devnani
AI/ML Engineer · GenAI · RAG · Speech AI
I build production-grade machine learning systems — from RAG architectures and speech intelligence pipelines to scalable ML infrastructure using Python and FastAPI. This blog is where I share deep technical breakdowns, experiments, and lessons from the field.
❯ whoami
Uday Devnani
❯ cat role.txt
AI/ML Engineer
❯ ls skills/
GenAI RAG STT
FastAPI Python
❯ echo $status
actively_writing ✓
❯
Featured
4-
Getting Started with Litmus for AI Model Testing: Ensuring Quality and Cost-Effectiveness
A comprehensive guide to using Litmus for AI model testing, evaluation, and optimization to ensure quality and cost-effectiveness in production.
-
Building Efficient AI Applications with LangChain: A Step-by-Step Guide
Learn how to build and deploy efficient AI applications using LangChain. A complete tutorial covering setup, integration, and deployment with practical examples.
-
Cleaning the Signal: Our 6-Step Audio Enhancement Pipeline using DeepFilterNet & FFmpeg
A practical 6-step audio enhancement pipeline using DeepFilterNet and FFmpeg to improve transcription accuracy.
-
Async at Scale: Orchestrating Audio Ingestion with FastAPI, Celery, and Redis
How we scaled asynchronous audio ingestion with FastAPI, Celery, and Redis for high-volume STT systems.
Recent Posts
5-
We Tested 8 STT APIs for Indian Languages So You Don't Have To
A benchmark of 8 STT providers for Indian language and Hinglish workflows with practical deployment guidance.
-
Fatal vs. Critical: Automating Risk Management for High-Stakes Sales Calls
Designing a severity-confidence framework to automate risk detection for high-stakes sales conversations.
-
The Cost of Bad Audio: How Pre-Processing Saves Us 20% on Transcription APIs
How silence removal and smart audio pre-processing reduced transcription API spend and improved output quality.
-
Generative AI for Rule Enforcement: structuring Prompts for Precision
How to engineer structured prompts for high-precision compliance auditing with LLMs.