Staff ML/AI Engineer – Test‑Time Verification & Canonical Manifold

corpora-jobs-1778796293285-db9d41c6 - Frontier Development

ML/AI EngineerStaff1 position

⚡

Why This Role is Different

Frontier Development Role

Lead the creation of a real‑time, self‑verifying inference stack that detects distribution shift and adversarial messages with sub‑5 ms latency, while producing transparent audit logs for human operators.

The Frontier Element

You will combine amortized latent steering, self‑supervised adaptation, and cross‑modal manifold alignment into a single, inference‑time module that operates without back‑propagation, pushing the boundary of real‑time AI safety.

🔬

Project Context

Research Area

Test‑Time Verification Layer (TTVL)

From: Theory of Mind Defenses Against Communication Sabotage

Why This Role is Critical

Build a lightweight, manifold‑aware verification module that flags anomalous messages in real time, enabling auditability and preserving cooperative performance.

What You Will Build

A canonical manifold learning pipeline, amortized latent steering module, and a low‑latency verification layer that integrates with the agent’s decision loop.

🛠

Key Responsibilities

Design and train a canonical manifold that captures the essential structure of legitimate communication patterns.
Implement amortized latent steering and self‑supervised adaptation for real‑time verification.
Integrate the TTVL into the agent’s message‑processing pipeline with <5 ms inference overhead.
Develop audit‑ready logging and visualization tools for human operators.
Validate the system against a suite of adversarial scenarios and publish results.

🎯

Required Skills & Experience

Technical Must-Haves

Manifold learning & dimensionality reduction

Expert

Building the canonical interaction manifold.

Self‑supervised learning (SAF, ALS)

Advanced

Enabling test‑time adaptation without back‑prop.

Real‑time inference engineering

Expert

Ensuring sub‑5 ms latency on edge hardware.

TensorFlow/PyTorch + C++ integration

Expert

Deploying models in production.

Cryptographic logging & tamper‑proof audit trails

Proficient

Enabling regulatory compliance.

Experience Requirements

5+ years in ML with a focus on real‑time systems.
2+ years in deploying low‑latency verification or anomaly detection modules.
Published work on TTVL, ALS, or SAF.

Education

PhD in Computer Science, Machine Learning, or related field.

⭐

Preferred Skills

Experience with cross‑modal consistency (vision‑language) verification.
Knowledge of SIEM integration for real‑time alerting.
Familiarity with GDPR‑style explainability requirements.

🤝

You Will Thrive Here If...

Action‑oriented, builds production‑ready modules from scratch.
Thrives in high‑autonomy, end‑to‑end ownership.
Enjoys rapid iteration and measurable impact.

📈

Impact & Growth

12-Month Impact

Deliver a TTVL that achieves <0.5 % false‑positive rate, <5 ms verification latency, and provides fully auditable logs, enabling the HTMAD system to meet stringent regulatory standards within 12 months.

Growth Opportunity

Lead the verification stack across all multi‑agent products, mentor a team of ML engineers, and shape the company’s AI safety and compliance roadmap.

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.