MS
Mahi's Profile

Mahi Sidda

ROLE: MACHINE LEARNING ENGINEER

LOCATION: TEMPE, AZ

EDUCATION: ARIZONA STATE UNIVERSITY

FOCUS: RAG · LLM INFERENCE · PRODUCTION AI

About Me

I'm an ML Engineer focused on building production-grade AI systems — from retrieval-augmented generation (RAG) pipelines to on-device LLM inference under real-world constraints. I enjoy working at the intersection of machine learning, systems, and performance optimization, translating research ideas into scalable, measurable products.

Recently, I've worked on LLM-powered document understanding, recommendation systems, and offline AI deployments, with an emphasis on latency, cost, and reliability.

Work Experience

I specialize in building end-to-end ML pipelines for real products — from data preprocessing and model selection to evaluation, deployment, and performance tuning. My work emphasizes practical impact: faster systems, better retrieval quality, and reliable production behavior.

Tools

I primarily work with Python-based ML stacks and modern LLM tooling, including PyTorch, Hugging Face Transformers, PEFT (QLoRA), scikit-learn, NumPy, and pandas. For production systems, I use FastAPI/Flask, Docker, Redis, PostgreSQL, FAISS, and Git/GitHub.

I also have hands-on experience deploying LLMs using llama.cpp, working with on-device inference, and optimizing systems for latency and memory constraints.

Here's what they say about me!

"Mahi is a thorough professional. He was part of our AI research + dev team and led a team of 3 to create a highly optimised RAG model from scratch, customised for our hardware architecture. The deliverables included prototyping, research, and model tweaks to adhere to cost structures, understanding business needs, and further incorporating changes from UX feedback with seamless collaboration. Would highly recommend, kudos!"

Arvind Mohan

- Arvind Mohan

Co-founder,CPO, hashmint

"Mahi laid the foundational architecture for AI at HashMint with the launch of MintPDF, our Retrieval-Augmented Generation (RAG) system that answers queries directly from user-uploaded documents. He engineered the entire backend pipeline, integrated OpenAI APIs, and built production-grade LLM workflows that made MintPDF fast, accurate, and reliable from day one. That foundation has been critical. Much of our rapid progress in AI since then has been possible because of the strong technical and architectural base Mahi set early on. He is a true 100x engineer who moves fast, solves hard problems, and quietly unlocks massive leverage for the entire team."

Yaswanth Rayapati

- Yaswanth Rayapati

Co-founder,CEO,hashmint

"Mahi is a thorough professional. He was part of our AI research + dev team and led a team of 3 to create a highly optimised RAG model from scratch, customised for our hardware architecture. The deliverables included prototyping, research, and model tweaks to adhere to cost structures, understanding business needs, and further incorporating changes from UX feedback with seamless collaboration. Would highly recommend, kudos!"

Arvind Mohan

- Arvind Mohan

Co-founder,CPO, hashmint

"Mahi laid the foundational architecture for AI at HashMint with the launch of MintPDF, our Retrieval-Augmented Generation (RAG) system that answers queries directly from user-uploaded documents. He engineered the entire backend pipeline, integrated OpenAI APIs, and built production-grade LLM workflows that made MintPDF fast, accurate, and reliable from day one. That foundation has been critical. Much of our rapid progress in AI since then has been possible because of the strong technical and architectural base Mahi set early on. He is a true 100x engineer who moves fast, solves hard problems, and quietly unlocks massive leverage for the entire team."

Yaswanth Rayapati

- Yaswanth Rayapati

Co-founder,CEO,hashmint