Artificial Intelligence Engineer

About Mvaak

Mvaak is an AI data services company specializing in egocentric (first-person) video data. We capture, process, and deliver real-world datasets that AI labs, robotics teams, and data partners use to train the next generation of multimodal and embodied AI models. We are a lean, fast-moving team where early-career engineers get real ownership from day one.

Role Overview

We are looking for an early-career AI Engineer who wants to learn fast and build hands-on. You will help develop the LLM, vision-language, and automation workflows that turn raw egocentric video into model-ready datasets. You will work alongside senior team members and grow quickly across GenAI, computer vision, and data engineering.

Key Responsibilities

• Build and improve pipelines for processing, captioning, and tagging egocentric video using LLMs and VLMs.
• Write scripts for automated quality checks (blur, frame drops, face/PII redaction, metadata validation).
• Help build AI agents and automations for dataset QA and packaging.
• Experiment with open-source vision and language models and evaluate their output.
• Support dataset preparation and delivery to client specifications.
• Document your work and share learnings with the team.

Requirements & Tech Stack

• Good Python skills.
• Basic understanding of machine learning and deep learning concepts.
• Hands-on experience with at least one of: PyTorch, Hugging Face, OpenCV.
• Has built at least one project using LLM APIs (Claude, GPT, Gemini) or open-source models.
• Curiosity, ownership, and a willingness to learn quickly.

Qualifications

• Bachelor’s or Master’s in Computer Science, AI, Data Science, or a related field (final-year students may apply).
• 0–2 years of experience, including internships, academic, or personal projects in AI/ML.

Good to Have

• Computer vision or video processing projects (OpenCV, FFmpeg).
• Exposure to LangChain, LangGraph, RAG, or vector databases.
• Basic FastAPI/Flask and cloud (AWS or GCP) experience.
• Familiarity with annotation tools (CVAT, Label Studio).
• GitHub portfolio, Kaggle work, or open-source contributions.