Sobre este puesto de Senior Applied ML Engineer en Higgsfield
Why work at Higgsfield AI?
Higgsfield AI is the fastest-scaling generative AI company in history, hitting $500M in annual revenue run rate, 25M+ users worldwide, 6M+ generations per day, and powering 390 of Fortune 500 brands.
We're building at the absolute frontier of AI-powered video creation and next-generation creative tools. Joining Higgsfield means becoming part of a high-impact team shaping the future of AI-native experiences, at a company that isn't just moving fast, but rewriting what fast looks like.
Why this role exists
Higgsfield is one of the fastest-growing generative AI companies in the world. At our scale, a weak classifier means legal exposure, refund losses, or a viral backlash. Applied ML sits between the model shelf, legal, and payments, and it's understaffed for the traffic we serve.
What you'll actually do
You'll own one of two tracks, decided at offer stage based on fit.
Product track — models in the generation pipeline:
• NSFW classification: detection, threshold calibration, appeals, adversarial resistance.
• IP and likeness classification across prompts and generated outputs.
• Anti-fraud: multi-account farms, free-trial abuse, refund abuse, prompt-injection cost exploits.
Data track — models and analyses that shape what we build:
• Churn and LTV prediction: user-level models that flag at-risk cohorts before they leave and forecast lifetime value across plans and geographies.
• Abuse and fraud clustering: unsupervised segmentation of account graphs, payment patterns, and generation behavior to surface new abuse vectors before rules can be written.
• Causal analysis of product changes: uplift modeling, quasi-experiments, and A/B test design when clean randomization isn't possible.
Either track is end-to-end. Define the problem, build the dataset, ship to prod, measure the dollar impact. Our release cycle is days, not quarters, so your first shipped model happens fast.
Must-haves
Shipped multiple ML systems to production at scale, actual serving traffic, not notebooks.
Deep experience in at least one of: computer vision / multimodal, content moderation / trust and safety, fraud or anomaly detection, recommendation or ranking.
You can define the problem yourself. We hand you an ambiguous business question, not a labeled dataset.
Strong Python and PyTorch. Comfortable with large-scale data infra.
Product judgment. You pick precision/recall thresholds and defend them in dollars, not F1.
Who this is not for
• You prefer working from complete specifications rather than defining problems yourself.
• You are looking for a slower release cycle. We ship weekly, and launches can be high-tempo.
Nice to have
• Trust and safety at a UGC or generative platform.
• Adversarial ML, you've fought people trying to bypass your classifier.
• Fine-tuning VLMs, embedding models, or perceptual hashing.
What we offer:
Competitive base salary in USD
Equity: participation in the company’s stock option program, giving you the opportunity to share in the company’s long-term growth.
On-site role in our Almaty office (we will relocate you from anywhere).
A seat where your classifier decision moves the P&L on the same day
