Machine Learning Engineer

VolumX is building the future of filmmaking. We work at the intersection of cinema, artificial intelligence, and high-end visual effects, creating advanced solutions for feature films, studios, brands, and celebrities. With over 40 film titles delivered, VolumX is trusted by leading filmmakers, celebrities, and Oscar-winning studios. We are looking for ML Engineers to help us build film-grade AI systems for face replacement, lip sync, neural rendering, de-aging, and digital human performance editing.

Role Overview As an ML Engineer at VolumX, you will work on next-generation AI systems for digital humans and cinematic facial performance. The role focuses on neural rendering, audio-driven facial animation, identity-preserving face synthesis, temporal consistency, and high-resolution video workflows. This is a hands-on role for someone who can read research papers, implement models, train systems on custom datasets, optimize inference, and turn AI research into reliable production tools.

Key Responsibilities

• Build AI models for neural rendering, lip sync, face replacement, de-aging, face reenactment, and digital human performance editing.
• Develop audio-driven facial animation and lip sync systems with accurate phoneme-to-viseme mapping, jaw motion, lip closure, teeth visibility, and emotional nuance.
• Build identity-preserving face synthesis pipelines that maintain facial structure, skin texture, pores, stubble, wrinkles, lighting, and micro-details.
• Improve temporal consistency across video frames to reduce flicker, jitter, identity drift, texture swimming, and unstable mouth regions.
• Work on face tracking, dense landmark tracking, 3D face reconstruction, mesh tracking, UV mapping, expression capture, and performance transfer.
• Train and fine-tune custom models on actor-specific datasets, multi-view footage, 4D capture, facial performance data, and production video footage.
• Develop tools for segmentation, occlusion handling, mouth interiors, hairline boundaries, skin masks, and seamless blending with original plates.
• Optimize training and inference pipelines for GPU efficiency, batch processing, cloud deployment, and artist-facing workflows.
• Collaborate with VFX artists to make AI outputs usable in Nuke, Maya, Unreal Engine, and other production tools.

Skills & Requirements

• Strong experience with Python, PyTorch, OpenCV, FFmpeg, NumPy, and GPU-based model training.
• Strong understanding of computer vision, deep learning, generative AI, and video processing.
• Experience with neural rendering, image/video generation, diffusion models, GANs, transformers, autoencoders, or 3D-aware generative systems.
• Experience with face-related AI tasks such as face detection, alignment, tracking, segmentation, reenactment, face swapping, or performance transfer.
• Understanding of facial geometry, landmarks, head pose, expression tracking, visemes, phonemes, and audio-driven facial motion.
• Experience with temporal video consistency, optical flow, motion estimation, frame coherence, video stabilization, and artifact reduction.
• Familiarity with 3D face models, mesh tracking, UV space, texture maps, blendshapes, facial rigging, FLAME, DECA, EMOCA, MediaPipe, Dlib, OpenFace, or similar tools.
• Experience with Gaussian splatting, NeRFs, COLMAP, 3D reconstruction, neural avatars, neural radiance fields, or volumetric rendering is a strong plus.
• Knowledge of loss functions such as perceptual loss, identity loss, landmark loss, adversarial loss, optical flow loss, temporal consistency loss, reconstruction loss, and photometric loss.
• Experience with segmentation, matting, face parsing, mouth segmentation, occlusion-aware masking, and plate integration.
• Ability to train and fine-tune models on custom identity-specific or actor-specific datasets.
• Experience with model optimization and deployment using ONNX, TensorRT, TorchScript, Docker, CUDA-aware workflows, or similar tools.
• Experience with multi-GPU training, distributed training, mixed precision, experiment tracking, and large-scale dataset handling is preferred.
• Exposure to Nuke, Maya, Blender, Unreal Engine, MetaHuman, KeenTools, Wrap3D/4D, or VFX production workflows is preferred.
• Ability to read research papers and convert them into working prototypes.

Ideal Candidate You love solving hard technical problems and want your work to matter. You are not looking for a routine 9 to 5 role, you want to be part of a team that is building something ambitious, difficult, and industry-changing. We value people who are sincere, trustworthy, loyal, and deeply committed to their craft. This role is for someone who can take ownership, push through challenges, stand by the team, and go above and beyond when the work demands it. You should be excited by the opportunity to build world-class technology from India for the global film industry. At VolumX, you will get to work closely with filmmakers, celebrities, directors, producers and technologists, while helping shape tools that can change how movies are made. We are looking for someone who wants responsibility, not just a job, someone who wants to grow with the company, contribute meaningfully, and help build best-in-class technology.

Why Join VolumX

• We believe in building the best. Over years we have built the most advanced technologies for Digital Humans and Film Making.
• We are trusted by top celebrities and filmmakers, you'll get a chance to work with them.
• Solve real problems that will be used to shape movies all over the globe.
• Join a team that values trust, loyalty, ownership, excellence, and long-term commitment.
• For exceptional people who prove themselves, we are open to meaningful long-term growth opportunities, including leadership responsibility and ownership participation.
• Be part of building a company from India that has the ambition to serve the global film and entertainment industry.

Location Gurgaon / Mumbai

How to Apply Please share your resume, GitHub, portfolio, research work, demos, or any relevant projects showing your experience with AI video, computer vision, neural rendering, lip sync, face replacement, 3D reconstruction, digital humans.