SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Mecka AI is building the data infrastructure layer for robotics and embodied AI, designing and operating global systems for data capture, data labeling, and hardware-enabled workflows used by leading AI labs and robotics companies to train and validate humanoid and embodied AI systems.
This role is a high-leverage force multiplier for the core Computer Vision Machine Learning (CVML) team. You will own the critical infrastructure, internal tooling, and targeted ML services that enable the CVML team to move fast, rather than building foundational models.
Key responsibilities include owning the internal annotation platform (CVA) used by the labeling team, shipping new task types, review workflows, keyboard-driven UI features, and progress dashboards. You will write data pipeline glue code to efficiently move video frames, metadata, and JSON payloads between cloud storage, databases, and client applications. You'll build tools to track inter-annotator agreement, audit sampling, and dataset health.
On the ML services side, you will own the PII blur pipeline that detects and blurs faces, screens, and license plates in delivered footage—a strict, customer-facing system balancing privacy compliance with data preservation. You'll deploy and optimize lightweight detection and segmentation models (e.g., bounding box assists) so annotators correct rather than create from scratch. You'll take off-the-shelf or provided PyTorch models, optimize them (e.g., TensorRT, ONNX), and wrap them in fast, concurrent APIs to serve the tooling UI.
Required skills include heavy frontend engineering (React, Vue, or modern JS/TS) with ability to handle complex state, canvas-based rendering, or video playback without performance degradation. You need production-grade Python expertise building fast, concurrent APIs (FastAPI, gRPC) that handle high-volume media requests. Practical applied ML experience is essential—fine-tuning and deploying detection, segmentation, or tracking models on messy, real-world data, and taking models from Jupyter notebooks into reliable pipelines.
Strong signals include building or heavily customizing annotation UIs (bounding boxes, polygons, keypoints), deep familiarity with video processing pipelines (FFmpeg, frame extraction, codec handling, latency optimization), familiarity with active learning workflows or mining hard negatives, and caring deeply about UI/UX measured by end-user workflow speed.
This role is not suitable for core ML architects focused on novel neural network architectures or frontier foundation models, backend-only engineers who dislike TypeScript/CSS/UI workflows, or engineers expecting perfectly clean, balanced datasets. You are building the systems that create and sanitize data.
