December 15, 202410 minComing Soon
Migrating from GPT-4 to GPT-3.5: A Cost Optimization Journey
How we achieved comparable accuracy at 5% of the original cost through careful prompt engineering and fine-tuning.
LLMCost OptimizationFine-tuning
Back to Home
Blog
Technical deep-dives, lessons learned, and thoughts on building production ML systems.
November 20, 202415 minComing Soon
Building a Two-Tower Recommendation System
From 70% to 90%+ match accuracy: lessons learned building ApplyPass's job matching engine.
EmbeddingsRecommendation SystemsML
October 5, 20248 minComing Soon
Real-time Speech Transcription with Whisper
Reducing latency from 30s to 5s in production call transcription.
Speech-to-TextWhisperFastAPI