constraints: noisy sensors, partial observability, physical laws, and high economic

stakes.

This is not offline research. You will own the full lifecycle; from theoretical formulation and

experimentation to real-time inference, uncertainty estimation, and continuous retraining in

production.

What You’ll Own

• Orbital’s foundational time-series modelling stack

• Physics-informed and probabilistic model design

• Uncertainty quantification and robustness under sensor faults

• Research → production translation for time-series models

• Benchmarking standards and validation protocols used across the company

Job Requirements Must-Have Qualifications

• PhD in Computer Science, Statistics, Applied Mathematics, Physics, or related field

• First-author publications in time-series modelling, forecasting, signal processing, or

physics-informed ML

Client Confiden+al

Client Confiden+al

• 3+ years of hands-on research experience in time-series or sequence modelling

• Strong foundation in:

o Deep Learning

o Probabilistic modelling

• Expert Python skills with production-grade PyTorch code

• Experience deploying ML models into real systems

How We Work

• Research is judged by production impact, not paper count

• We value principled models that survive contact with reality

• We iterate aggressively, benchmark honestly, and ship responsibly

• Physics, statistics, and learning are treated as complementary, not competing

What This Role Is Not

• Not offline academic research disconnected from deployment

• Not pure deep-learning experimentation without domain grounding

• Not feature engineering on static datasets

• Not a support role; this position owns core IP

Job Responsibilities

1. Design & Implement Foundational Time Series Models

• Design core time-series architectures supporting:

o Forecasting

o Classification / anomaly detection

o Optimisation & control-adjacent tasks

• Explore and select appropriate objectives examples:

o Probabilistic losses

o Generative formulations

o Reinforcement-learning-inspired objectives where appropriate

• Develop hybrid approaches that blend:

o Classical statistical models

o Deep learning architectures

o Physics-based constraints

• Improve generalisation, interpretability, and extrapolation beyond training regimes

• Ensure models respect physical feasibility in production settings

3. Uncertainty, Robustness & Reliability

• Design uncertainty-aware models (Bayesian, ensemble, hybrid)

• Quantify confidence under:

o Sensor drift and failure

o Regime change

o Sparse or delayed ground truth

• Ensure outputs are usable by operations and engineering teams, not just statistically

elegant

4. Production Structured AI Code

• Containerise and deploy models using Docker on AWS / Azure (EKS, ECS, SageMaker)

• Build or integrate CI/CD pipelines for:

o Training

o Evaluation

o Rollout and rollback

o Automated retraining triggers

5. Benchmarking & Validation

• Define rigorous back-testing and evaluation protocols

• Build automated benchmarking pipelines across datasets, regimes, and failure modes

• Compare against classical baselines and modern deep-learning approaches

• Ensure claims are defensible to customers, partners, and internal stakeholders

2. Embed Physics-Informed Structure

• Integrate domain physics into learning systems, including:

o Conservation laws

o Process constraints

o Differential-equation-based priors

What Success Looks Like

First 90 Days

• Deep understanding of Orbital’s data, domains, and production constraints

• Contribution to at least one core time-series model or experimental track

• Clear ownership of a modelling problem with defined success metrics

6–12 Months

• One or more foundational models running reliably in production

• Demonstrable improvements in:

o Forecast accuracy

o Robustness under faults

o Uncertainty calibration

• Models actively used by downstream agents and optimisation layers

• Benchmarking standards adopted across the research team

Job Benefits

• Remote or hybrid role with an office in Fitzrovia

• Competitive salary

• Attractive set of benefits