Staff Machine Learning Engineer
Full Job Description
As a Staff Machine Learning Engineer focused on Video Insights at JioStar, you will leverage your strong technical independence and collaborative spirit to deliver high-quality projects. You will act as a functional lead within your domain, guiding and mentoring team members. Join our Video Center of Excellence (CoE) team to fine-tune and optimize large video models using cutting-edge multimodal AI. You will experiment with state-of-the-art video understanding models, customize them for real-world applications, and contribute to a product that impacts millions of users by enhancing video insights and content comprehension. This role offers the opportunity to tackle complex ML challenges at scale within an entrepreneurial environment, combining deep ML expertise with robust deployment infrastructure.
Key Responsibilities:
- Fine-tune large video models (Vid-LLMs) using advanced techniques like LoRA, QLoRA, and PEFT for specific video understanding tasks.
- Design and implement efficient model adaptation pipelines tailored for domain-specific video content and use cases.
- Optimize model inference performance through quantization, knowledge distillation, and hardware-specific optimizations.
- Conduct extensive experimentation and ablation studies to determine optimal model configurations and hyperparameters.
- Build robust evaluation frameworks and metrics to assess model quality, generalization capabilities, and performance on edge cases.
- Collaborate with research and product teams to translate business requirements into actionable model tuning objectives.
- Develop and maintain comprehensive documentation for tuning methodologies, lessons learned, and team best practices.
- Contribute to open-source projects and stay abreast of the latest advancements in multimodal AI and video understanding.
Skills and Attributes for Success:
- Over 6 years of professional experience in machine learning engineering, with a specialized focus on deep learning and model fine-tuning.
- Advanced proficiency in Python and hands-on experience with deep learning frameworks, particularly PyTorch.
- Proven experience in fine-tuning large language models and multimodal models using techniques such as PEFT, LoRA, and similar approaches.
- Strong understanding of video codecs, video processing pipelines, and streaming technologies.
- Solid foundation in computer vision and deep learning fundamentals, including CNNs, Transformers, and attention mechanisms.
- Experience with model evaluation frameworks, A/B testing, and continuous experimentation infrastructure.
- Proficiency with GPU-based training and inference optimization using CUDA or equivalent frameworks.
- Excellent problem-solving abilities and capacity to debug complex ML systems in production environments.
- Familiarity with version control systems (Git) and MLOps tools like MLflow, Weights & Biases, or similar platforms.
Preferred Education and Experience:
- Bachelor of Engineering/Technology (BE/B.Tech) in Computer Science, Electrical Engineering, AI, or a related technical field, with 7 to 9 years of experience.
- A Master of Science (MS) or Doctor of Philosophy (PhD) in Machine Learning/Artificial Intelligence is a plus.
Company
JioStar
JioStar is a premier global media and entertainment company dedicated to revolutionizing how audiences engage with content, particularly in entertainment and sports. With a vast television network and...