Computer Vision Engineer

Pangram Labs•New York, NY

1d•$230,000 - $260,000•Onsite

About The Position

Pangram Labs is hiring a Computer Vision Engineer to build and ship the multimodal AI detection systems at the core of our product. You'll own models end to end — from data and training through deployment, serving, and monitoring in production — to build the most accurate AI detection software in the industry. You'll train deep learning models that reliably identify AI-generated images, then take responsibility for getting them into production and keeping them performant, reliable, and continuously improving against an adversarial, fast-moving threat landscape. This is an engineering-first role: research is part of the work, but the job is measured by what ships and how well it runs at scale.

Requirements

3–5+ years of experience building and deploying ML systems in production, and/or a Ph.D. in AI, Machine Learning, Computer Vision, or a related area
A strong engineering track record: shipping production code, deploying models, and building reliable systems. This can be demonstrated through significant contributions to AI products in industry, or through first-author publications at major AI/CV venues (CVPR, ICCV, ECCV, NeurIPS, ICML, ICLR, etc.)
Experience training neural networks in PyTorch at scale, including distributed training
Experience with AI infrastructure and distributed systems — model serving, pipelines, and the tooling required to run models in production
Deep understanding of generative image models (diffusion models, GANs, autoregressive image models) and how they work
Solid understanding of neural networks for video, audio, time series, and signal processing
Familiarity with recent technical developments in LLMs

Nice To Haves

Experience with MLOps tooling, CI/CD for ML, or model observability and monitoring at scale
Experience or publications in image forensics or deepfake detection
Experience with synthetic data, adversarial robustness, or interpretability

Responsibilities

Train large-scale deep learning models to detect AI-generated content, and own the full lifecycle: data pipelines, training, evaluation, deployment, monitoring, and retraining
Build and operate the production systems that serve detection models — inference services, batch pipelines, and the infrastructure behind them — with attention to latency, throughput, cost, and reliability
Design evaluation and monitoring pipelines that catch model regressions and detect drift as new generative models emerge
Build infrastructure for synthetic dataset creation and large-scale data processing
Improve model accuracy, robustness, and efficiency through experimentation grounded in production metrics
Work with engineering and product teams to translate models into shipped product features, and share results with technical and non-technical audiences through documentation, whitepapers, and blog posts