Experience: 2–8 years

Location: Work from home initially

Role summary

AI Standards is building sovereign AI systems for enterprise environments, with work spanning AI safety, cybersecurity, quantum defence and on-premises deployment. We’re looking for an AI/ML Engineer with strong mathematical foundations who can help take models from experimentation to reliable deployment.

You’ll work closely with the founding and engineering teams on model development, fine-tuning, evaluation and inference optimisation. Your work will contribute to systems designed for Fortune 1000 requirements, where performance, privacy, traceability and operational reliability matter together.

Responsibilities

- Translate product and research objectives into well-defined ML problems, measurable baselines and reproducible experiments
- Evaluate and select suitable models based on task quality, licensing, hardware requirements, latency and deployment constraints
- Build data preparation pipelines covering cleaning, validation, deduplication, dataset versioning and leakage prevention
- Develop and fine-tune models, including parameter-efficient adaptation for customer-specific domains and datasets
- Design evaluation suites that test accuracy, robustness, generalisation, safety and failure modes, rather than relying on a single benchmark
- Analyse model errors and use evidence to decide whether improvements require better data, training changes,



retrieval, routing or a different model
- Optimise inference through appropriate techniques such as quantisation, batching, caching and efficient serving, while measuring quality trade-offs
- Profile GPU utilisation, memory consumption, throughput and latency to identify bottlenecks under realistic workloads
- Collaborate on model routing and orchestration, including confidence thresholds, escalation paths and specialist-model selection
- Package models and their dependencies for repeatable, containerised deployment in customer-controlled environments
- Work with software and infrastructure engineers on model APIs, monitoring, versioning, rollback and deployment validation
- Contribute to adversarial testing, sensitive-data protection and safeguards against misuse or unreliable model behaviour
- Maintain explicit experiment records, model documentation and deployment instructions so systems can be operated without depending on one individual
- Review relevant research and test whether new approaches deliver measurable value within our existing architecture

What success looks like You deliver models whose improvements are demonstrated through repeatable evaluations. Deployments meet agreed quality and resource targets, limitations are clearly documented, and other engineers can reproduce, maintain and extend your work.

📌 AI/ML Operations Engineer and Strategist (Kolkata)
🏢 AI Standards
📍 Kolkata