class SmeetMehta:
def __init__(self):
self.role = ["Backend Engineer", "AI/ML Engineer"]
self.education = "MS @ USC (Applied Data Science)"
self.languages = ["Python", "Java", "C++", "SQL"]
self.backend = ["REST APIs", "PostgreSQL", "Docker", "Linux", "Distributed Systems"]
self.ai_ml = ["LangChain", "Multi-Agent Systems", "RAG", "LLMs", "PySpark"]
self.currently = "Building AI systems that work in production - not just demos"
self.work_auth = "F-1 OPT - Authorized to work in the US"
def __str__(self):
return "Engineers clean backends. Deploys smarter AI. Ships both."🏥 Clinical AI Platform - Lumina AI Health
Designed and deployed a PostgreSQL-backed multi-modal backend ingesting lab results, imaging & genomic data. Engineered 8 production LangChain agents for automated clinical recommendations via Firebase Studio.
🤖 Insights-Pro - Tiger Analytics
Owned end-to-end architecture of a multi-agent AI system - Web Search + SQL Generation + Validation agents - hitting 90% accuracy on NL business queries with 80% boost in GenAI output efficiency.
💰 Price Optimization Engine - Tiger Analytics
Built a derivative-free constrained optimization model in Python delivering a 30% average profit increase for a multinational food giant across 10+ markets.
♻️ Large-Scale Codebase Refactor - Tiger Analytics
Modularized a 100K+ line production codebase into region-configurable architecture - cut production incidents by 40% and slashed regional rollout time by weeks.
Backend & Systems
AI / ML / GenAI
| Project | What it does | Stack |
|---|---|---|
| 🍽️ Yelp Recommender | Hybrid recommendation engine - 0.972 RMSE on 1.2GB+ data | PySpark, XGBoost |
| 🌍 LinguaSense | Real-time multilingual sentiment with NL explanations | HuggingFace, LLMs |
| 🗄️ ConvoDB | NL → SQL/MongoDB query engine, zero LLM, built from scratch | NLTK, Python, SQL |
I'm actively looking for Software / Backend / AI-ML Engineering roles. If you're building something hard and meaningful - I'd love to hear about it.
