Senior Generative Observation Modeling Engineer

corpora-jobs-1778796293285-db9d41c6 - Frontier Development

Research EngineerSenior1 position

⚡

Why This Role is Different

Frontier Development Role

Lead the cutting‑edge development of conditional generative models that can reconstruct corrupted multimodal sensor data in real time, pushing the limits of GAN stability and privacy on distributed agents.

The Frontier Element

You’ll design a hybrid GAN that integrates physics‑based loss terms and differential privacy into a lightweight architecture that can run on UAV swarms, a first in the field of adversarial observation inference.

🔬

Project Context

Research Area

Generative Observation Modeling (CC‑GAN) for reconstructing missing or corrupted sensor streams

From: Adversarial Observation Perturbations and Policy Inference

Why This Role is Critical

Critical to build and maintain the conditional GAN that learns the joint distribution of clean and perturbed observations, enabling real‑time reconstruction and anomaly detection for multi‑agent fleets.

What You Will Build

Offline training pipeline for CC‑GAN, physics‑based regularizers, differential‑privacy‑aware federated training module, and a lightweight inference engine for edge UAV hardware.

🛠

Key Responsibilities

Design and implement the CC‑GAN architecture, incorporating multimodal conditioning and physics‑based regularizers.
Develop a federated training pipeline that preserves privacy while aggregating gradients from edge devices.
Engineer a real‑time inference engine that reconstructs corrupted sensor streams on low‑power UAV hardware.
Validate reconstruction quality using downstream task performance and physical sensor model consistency.
Iterate on training stability, mode collapse mitigation, and model compression for deployment.

🎯

Required Skills & Experience

Technical Must-Haves

Conditional GANs (e.g., Pix2Pix, CycleGAN) with multimodal conditioning

Expert

Core to learning joint distribution of clean and perturbed observations.

Physics‑based loss design and domain‑specific regularizers

Advanced

Ensures realistic reconstructions that respect sensor physics.

Federated learning and differential privacy

Advanced

Critical for secure, privacy‑preserving training on distributed UAV fleets.

Edge ML deployment (TensorRT, ONNX, TinyML)

Proficient

Required to run the generative model on resource‑constrained hardware.

Python, PyTorch / TensorFlow, CUDA programming

Expert

Daily stack for model development and experimentation.

Experience Requirements

5+ years in generative modeling or computer vision research.
Track record of publishing in top ML conferences (NeurIPS, ICML, CVPR).
Experience deploying ML models on edge devices or UAV platforms.

Education

PhD in Machine Learning, Computer Vision, or Robotics with specialization in generative modeling or privacy‑preserving ML.

⭐

Preferred Skills

Experience with diffusion models or latent‑variable generative models.
Knowledge of quantum‑enhanced digital twin concepts.
Background in sensor fusion and multimodal data processing.

🤝

You Will Thrive Here If...

Demonstrated ability to ship end‑to‑end systems in fast‑paced environments.
Comfort with high‑risk experimentation and rapid iteration.
A passion for pushing the boundaries of what generative models can achieve.

📈

Impact & Growth

12-Month Impact

Within 12 months, deliver a production‑ready CC‑GAN module that reduces observation error by >30% on benchmark UAV datasets and supports federated inference on edge hardware, directly improving cooperative mission success.

Growth Opportunity

Lead a cross‑disciplinary team to extend the generative model to quantum‑enhanced digital twins and to integrate with the cooperative resilience layer, scaling the approach to large‑scale, heterogeneous fleets.

Ready to Push the Boundaries?

If this sounds like the challenge you have been looking for, we want to hear from you. We value what you can build over where you have been.