AI and ML Engineer

Location:  Mumbai, with ocassional travel to customer sites 

Type:  Full-time 

Experience:  3+ years, hands-on 

Reports to:  CTO

Role Overview 

SmartBeam AI builds physics-informed machine learning for industrial rotating assets. We combine IIoT, sensor telemetry, edge hardware and physics models to predict failures and improve energy and reliability metrics for mission-critical industrial companies like data centers, refining & chemicals, power plants and others. 

We are hiring an AI Engineer to own our data, models, deployment and AI architecture. This is the technical seat next to the CTO. You decide how data is curated, how models are designed and validated, how they reach production, and how the pieces fit together as a system. Research rigour and production delivery matter equally here. You will defend a modelling choice to a technical reviewer and work closely with our research partner (IIT Roorkee) on a daily basis. 

Key Responsibilities 

• Own the modelling lifecycle: data preparation, architecture selection, training, testing, validation, versioning and retraining. 

• Design, train and test physics ML models for anomaly detection, fault classification, degradation profiling and remaining-life estimation across IIoT telemetry channels. 

• Research and implement new methods such as deep neural networks, or new applications of existing ones, to improve detection sensitivity and reduce false alarms. 

• Work with the CTO on the physics-informed layer: calibrated first-principles outputs and residuals into data-driven diagnosis, and physics formulation into trainable, documented code. 

• Own the AI architecture and build the MLOps pipelines and cloud services for model serving, drift detection and monitoring. 

• Own CI/CD workflows using GitHub Actions and MLOps, covering build, test, model packaging and release. 

• Develop containerised solutions (Docker) and orchestrate deployments using Kubernetes, AKS or EKS with a strong focus on performance, reliability and observability. 

• Apply robust security controls, identity and access management, and cloud governance practices aligned with customer compliance requirements. 

• Troubleshoot and resolve infrastructure issues, deployment failures, performance bottlenecks, and system reliability concerns across the cloud stack. 

• Optimise models for low-latency inference on cloud, gateway and edge hardware. 

• Run data capture campaigns with the AI Lab and field teams: measurement plan, operating states, seeded-fault experiment design, ground truth and acceptance standards. 

• Maintain architecture documentation, deployment guidelines, engineering playbooks and reusable templates to standardise best practices. 

• Mentor the junior data scientist and interns, participate in design and code reviews, and carry the technical narrative in customer reviews. 

What You Need 

• Bachelor’s or Master’s from a tier-1 engineering or science institute, such as IIT, IISc, NIT, BITS Pilani or a comparable global institution, in Mechanical, Electrical, Instrumentation, Aerospace, Computer Science, Applied Mathematics or Physics. 

• 3 or more years building and shipping machine learning on real-world data or sensor time-series data. 

• High ownership. You have carried a system end to end without a support structure behind you, preferably in a startup, and shipped it to a paying or pilot user. 

• Strong machine learning fundamentals: architecture, training dynamics, regularisation and honest evaluation. You can explain why a model behaves the way it does, not only that it scores well. 

• Expertise in NumPy, Pandas, TensorFlow, SciKit Learn, PyTorch, Keras or equivalent, including custom training loops and working knowledge of convolutional networks, variational autoencoders and transformers. 

• Anomaly detection under scarce or noisy labels, with an approach to validation that does not leak. 

• Production Python and SQL, cloud engineering on AWS or Azure, Docker, Kubernetes, CI/CD and experiment tracking. Reviewed, version-controlled code rather than notebooks. 

• Physics literacy. Either physics-informed or hybrid modelling, such as PINNs, residual modelling, first-principles calibration or digital twins, or strong mechanical and thermo-fluid fundamentals you can apply to rotating and reciprocating equipment. 

Good to Have 

• Domain knowledge of industrial automation, communication and security protocols, and specific machine failure modes such as cavitation, bearing degradation, etc.  

• Edge inference tooling: ONNX, TensorRT, pruning or quantisation-aware training. 

• Time-series platforms and industrial protocols: Grafana, InfluxDB, TimescaleDB, Modbus, OPC-UA, MQTT. 

• Publications, patents or open-source contributions in applied ML, signal processing, spectral analysis, and handling of non-stationary and unevenly sampled time series data 

Why Join Us 

Nobody has yet made physics-informed AI work for industrial machines. At SmartBeam AI, you’ll have the autonomy to build it, break it and fix it, alongside a team that brings curiosity and rigour to every problem. You’ll work closely with the customers and operators who depend on what you ship, and with teammates who have your back. You’ll learn faster than you thought you could, be held to a high bar, and be paid competitively with performance-linked upside. If that sounds like your kind of place, join us. 

SmartBeam AI is an equal opportunity employer. We hire on evidence of clarity, capability and judgement. If this role is not the right fit but you see another way to contribute, write to us at careers@smartbeam-ai.com.