About Mecka AI

Mecka AI is building the data infrastructure layer for robotics and embodied AI.

We partner with leading AI labs and robotics companies to deliver high-quality, real-world datasets used to train, evaluate, and deploy robotic systems. Our work sits directly between research, data, and real-world execution - where model performance is dictated by data quality.

The Role

We are looking for a Machine Learning Engineer who operates from first principles. You aren't just an "applied" practitioner who treats models as black boxes; you are someone who understands the mathematical foundations of deep learning and has a sharp intuition for how architectural changes affect gradient flow, convergence, and generalization.

In this role, you are the technical connective tissue of the company. You will be the person everyone goes to for help with model development-from the math of the architecture to the reality of deployment-and you will apply your ML expertise to assist and elevate every engineering and research function at Mecka. You must be motivated by deadlines, good with debugging, and driven to see a product reach the hands of customers.

What You'll Work On Full-Cycle Model Development & Shipping Production-Grade Code: You will write, test, and debug high-performance production code. This isn't just about training scripts; it's about building robust, deployable systems. Deadlines & Delivery: We operate in a fast-paced environment. You should be motivated by clear milestones and the drive to ship a functional product on schedule. Fundamental Model Development & Intuition Architectural Rigor: Design and implement state-of-the-art models (Transformers, Diffusion models, etc.). You should understand exactly how changes in attention mechanisms, embedding spaces, or normalization layers impact performance. Mathematical Optimization: Derive custom loss functions and optimization strategies that capture unique constraints, ensuring the math supports the target behavior. Cross-Functional ML Advisory Engineering Support: Assist the platform and data engineering teams in building pipelines that are "ML-aware." You will help design how data is tokenized, compressed, and augmented to ensure zero loss in model-ready signal. Infrastructure Optimization: Use your knowledge of GPU kernels and distributed training to ensure our internal infrastructure is as efficient as the models themselves. Pipeline Innovation The Feedback Loop: Apply insights from model failures to improve our data processing. If a model fails due to a specific data distribution shift, you are the person who designs the automated filter or labeling tweak to fix it at the source. Who You Are

Required Background: Deep Theoretical Foundation: A degree (MS or PhD preferred) in a quantitative field (CS, Math, Physics, or Stats). You should be able to discuss the math behind backpropagation, scaling laws, and optimization algorithms from first principles. Production Mindset: You are an expert at writing and debugging clean, efficient code. You know how to take a research concept and harden it for a production environment. Fundamental Understanding: You know how models work at a low level. You can explain the "why" behind specific architecture choices and how they relate to the underlying mathematics. Hands-on Implementation: Mastery of PyTorch or JAX . You are comfortable writing custom training loops and debugging at the tensor level. Strong Signals: Deadline-Driven: You thrive when there is a clear product goal and a timeline to meet. Generalist Problem Solver: You can apply ML to a variety of domains, understanding the nuances of different data modalities. Cross-Disciplinary Mindset: You enjoy helping others and can translate complex ML concepts into actionable advice for data engineers and operations teams. Builder Mentality: You aren't afraid of a non-functional system. You find satisfaction in the "full-stack" journey of making a model move from a concept to a high-performing deployment. Why Mecka AI? High Agency: You are the authority on how we build and evaluate models. Core Influence: Your ML expertise will touch every part of our tech stack, from data capture to final deployment. Unrivaled Data: Work with the largest and highest-quality real-world datasets in the industry.