Sr. Machine Learning Engineer

Deep Learning Research Engineer – Computer Vision & Retail AI

Eizen AI | Hyderabad, India (R&D Hub) | Full-time | ESOPs

Fast-track your application: https://forms.office.com/r/y9JFDDEZqX

About This Role

Eizen AI is building the next generation of visual intelligence for retail. You will own deep learning models end to end: from reading the paper and designing the experiment to optimised inference running in production on store cameras, warehouse systems, and digital commerce platforms.

This is a research role with a production centre of gravity. Our models see what happens on the shelf, in the warehouse, and in the online catalogue, and turn it into merchandising, pricing, inventory, and supply-chain decisions for Fortune 500 retailers. That means real-world problems: thousands of near-identical SKUs, cluttered and occluded shelves, varied lighting across hundreds of stores, long-tail events with few labels, and strict latency budgets on edge hardware. Accuracy must translate into revenue, margin, and on-shelf availability, not just a benchmark score.

We combine deep learning with knowledge-driven reasoning and cognitive models. We are looking for an engineer-scientist who understands why algorithms work, connects that understanding to business outcomes, owns models from idea to production, and publishes what they learn.

What You'll Do

Retail AI

• In-store: shelf monitoring, out-of-stock detection, planogram compliance, fine-grained SKU and price-tag recognition.
• Store operations: footfall, dwell time, heatmaps, queue and checkout analytics, shrink and loss prevention, with privacy by default.
• Merchandising and pricing: connect visual signals to demand forecasting, assortment, and price and promotion optimisation.
• Supply chain: inventory counting, pallet and package recognition, damage detection, and fulfilment monitoring.
• Digital commerce: vision-language models for attribute extraction, catalogue enrichment, visual search, and recommendations.
• Turn retail questions into rigorous ML formulations and measure impact in sales lift, margin, availability, and shrink.Computer Vision and Foundation Models
• Build state-of-the-art models for detection, action recognition, temporal localisation, and multi-object tracking using video transformers (VideoMAE, TimeSformer, ViViT), SlowFast, and X3D.
• Fine-tune VLMs, multimodal LLMs, and vision foundation models (CLIP, SigLIP, DINOv2, SAM) with the Hugging Face ecosystem (Transformers, PEFT/LoRA, Accelerate, Diffusers).
• Apply self-supervised, contrastive, and few-shot learning, plus synthetic data, for scarce labels and new SKUs.
• Design evaluation that reflects deployment: per-store performance, domain shift, and failure-mode analysis.Graph Neural Networks and Time Series
• Apply GNNs to product-store-customer graphs, scene graphs, and human-object interaction.
• Build time-series models for demand forecasting, promotion effects, and anomaly detection.Training, Optimisation and Production
• Train at scale with mixed precision, DDP, FSDP, or DeepSpeed.
• Optimise inference with ONNX, TensorRT, quantisation, and distillation; deploy on NVIDIA GPUs and Jetson with DeepStream and Triton.
• Build model-serving APIs with FastAPI, and data pipelines for video ingestion, annotation, and versioning.
• Keep work reproducible with MLflow or Weights & Biases, tested with Pytest and CI, and shipped with Docker and Kubernetes.Research and Open Source
• Publish at NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, and AAAI.
• Contribute models, datasets, and tools to Hugging Face and open source.
• Represent Eizen in hackathons, challenges, and research communities.Must-Have Skills
• 2 to 5 years of hands-on deep learning experience, or an equivalent MS/PhD research record, with models trained and shipped or published.
• Deep understanding of the mathematics behind ML (optimisation, probability, linear algebra) and first-principles reasoning.
• Ability to connect algorithms to business value and explain trade-offs to non-ML stakeholders.
• Strong Python and PyTorch, including custom models, losses, and training loops.
• Hands-on computer vision and Transformer experience (ViT, DETR, video transformers, or VLMs).
• Practical fine-tuning with Hugging Face.
• Solid ML fundamentals: representation learning, training regimes, evaluation, and ablation design.
• GPU training workflows, CUDA basics, and mixed-precision or distributed training.
• Python services with FastAPI or Flask.
• Strong CS fundamentals: data structures, algorithms, complexity, and object-oriented design.
• Linux, Git, Docker, documentation, and unit testing.Nice to Have
• Publications at NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, AAAI, KDD, or TPAMI/IJCV.
• Kaggle (Expert and above), Smart India Hackathon, CVPR/ECCV workshop challenge, or similar wins.
• Open-source contributions to Hugging Face, PyTorch, OpenMMLab, Ultralytics, or your own well-used projects.
• Retail, CPG, or e-commerce AI experience: shelf analytics, product recognition, forecasting, pricing, or recommendations.
• Reinforcement learning or causal inference for pricing and inventory decisions.
• TensorRT, DeepStream, Triton, and Jetson deployment.
• Robotics (ROS 2), 3D vision, or agentic AI exposure.
• MLOps and cloud platforms (AWS, Azure, or GCP).Education

Bachelor's, Master's, or PhD in Computer Science, AI/ML, Electrical Engineering, Mathematics, Statistics, or a related field. Publications, competition results, and open-source impact are weighed as heavily as years of experience.

Why Join Eizen

• Retail AI at enterprise scale: your models shape merchandising, pricing, and supply-chain decisions for Fortune 500 retailers.
• Research that ships: production systems, not notebooks.
• Publish, publish, publish: time, compute, and support for top-conference submissions.
• A distinctive approach: deep learning plus cognitive reasoning, led by a founder with a PhD in Cognitive Robotics and 18+ years in AI.
• Ownership from day one in a small, senior team.
• ESOPs for high performers.Our Values

Put employees first. Think big. Aim for excellence. Own it and get it done. Embrace each other's differences.

How to Apply

Fill up the form for a faster response: https://forms.office.com/r/y9JFDDEZqX

Or send your CV to careers@eizen.ai

Learn more at https://eizen.ai

Eizen AI is an equal opportunity employer. We evaluate applicants on capability and potential, and welcome candidates from all backgrounds.