Navya Mehta
Early Engineer - ML Compilers at EnCharge AI
Old Toronto, Ontario, Canada
Summary
🤩
Rockstar🎓
Top SchoolNavya Mehta is an early-stage ML compiler engineer with 7 years of experience building high-performance systems spanning edge AI, backend infrastructure, and production ML. Currently an early software-side hire at EnCharge AI, she focuses on compiler infrastructure, quantization, and end-to-end simulation for a >150 TOPs/W edge chip. Previously a founding engineer at WOMBO.ai, she scaled async queuing, custom schedulers, and a <100ms vector-embedding semantic search to support 50M+ downloads and 1M DAU, blending low-latency systems engineering with ML deployment. Her background includes MLIR and C++ compiler work at Groq, ultra-low-latency trading systems at Jane Street, and data-driven modeling across finance and retail, showing a rare mix of algorithmic rigor and product impact. A Waterloo CS + Finance graduate and former research assistant in statistical optimization, she brings both theoretical depth and practical chops in optimization, memory-locality, and instruction compression. Notably, she has shipped novel compiler rewriters and graph-folding techniques that materially reduced IR size and instruction count on real-world networks.
7 years of coding experience
6 years of employment as a software developer
Grade 8-12 (IGCSE + IB), High School GPA: 3.99/4.00; IB World Topper: 45/45, Grade 8-12 (IGCSE + IB), High School GPA: 3.99/4.00; IB World Topper: 45/45 at Dhirubhai Ambani International SchoolBachelors of Honours Computer Science + Finance (Minor: Statistics), Grade: 94.18%, Bachelors of Honours Computer Science + Finance (Minor: Statistics), Grade: 94.18% at University of Waterloo
English, French, Gujarati
Github Skills (117)
fastapi10
python10
json-schema10
annotations10
swagger10
asgi10
openapi10
async10
django10
mypy10
swagger-ui10
api10
rest10
json10
redoc10
Programming languages (6)
Github contributions (5)
navyamehta/data-sentiment-NLP
Feb 2020 - Jan 2021
This project attempts to use a mix of traditional NLP techniques including trigrams, HNNs, and SVDs with attempts at LSTM and BERT architectures to predict and analyze sentiment projects - macroeconomic news relevance and Twitter subtext extraction.
Contributions:56 pushes, 1 branch in 11 months
extractionbertpredictsentiment-analysisnews
navyamehta/ai-research-projects
Sep 2020 - Feb 2021
Contributions:40 pushes, 1 branch in 5 months
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.