ML Engineer

Karan Mohindroo

I build tools that measure what quantization breaks.

scroll

Work

Featured Projects

Cross-backend differential testing CLI for LLM inference correctness — quantization sweeps, determinism verification, and serving-layer faithfulness across mlx-lm, vllm-mlx, and llama.cpp.

PythonMLXvLLMllama.cpp+1

Active

Native macOS app that adds three-finger trackpad gestures for middle-click and middle-drag. Distributed via Homebrew and MacPorts.

SwiftmacOSMultitouchSupportAccessibility API+2

Complete

Confidence-based filter decoding system that routes LLM inference requests between small and large models to reduce compute costs while maintaining output quality. BayLearn 2024 poster at Apple.

PythonPyTorchLLaMASpeculative Decoding+1

View all projects →

About

Who I am

ML Inference Engineer

Karan Mohindroo

CS & Behavioral Neuroscience at Northeastern. I build tools that test what quantization breaks — and find the optimal precision that doesn't.

Location

Oakland, CA

Looking for inference engineering roles

Currently building

infer-check

Cross-backend differential testing for LLM inference correctness

Stack

PythonSwiftMLXvLLMllama.cppPyTorchHomebrewPyPI

Open source

Contributor to Homebrew and MacPorts.

Research

~ BayLearn 2024 poster at Apple (CALID).
~ First published quant data for Qwen3.5-4B on MLX.

Writing

Latest from the Blog

I Built the LLM Inference Correctness Tool That Should Already Exist
Every benchmark measures tokens per second. Nobody measures whether the tokens are correct. I built infer-check to fix that — and ran it across Llama-3.1-8B and Qwen3.5-4B on Apple Silicon.
8 min read
Mar 2026

Read all posts →

Karan Mohindroo

Featured Projects

infer-check

MiddleDrag

CALID

Who I am

Latest from the Blog