Role Overview
We’re seeking a highly skilled ML Engineer – NLP Specialist with deep expertise in transformer models, real-time text analysis, and multilingual NLP. In this role, you’ll design and optimize machine learning systems that power intelligent, fraud-aware language understanding in chat environments. You’ll contribute to next-generation AI products that prioritize speed, context awareness, and real-world fraud defense, helping shape scalable solutions for millions of users across messaging platforms.
Key Responsibilities
- Build lightweight NLP models for real-time fraud detection.
- Design pipelines that detect social engineering patterns, scam links, urgency-based manipulation, and impersonation attempts.
- Train classifiers to detect masked fraud and microtext manipulation.
- Ensure real-time inference performance by optimizing models for edge devices.
- Integrate the model with mobile messaging environments (iOS/Android) and backend APIs.
- Collaborate with speech engineers, backend engineers, and AI PMs to align chat logic with broader fraud detection goals.
- Continuously improve fraud detection accuracy via adversarial training, feedback loops, and real-world test cases.
Required Technical Skills
- Excellent command of Python, PyTorch or TensorFlow, and text processing libraries.
- Ability to build end-to-end inference-ready pipelines that can run in mobile, edge, or server environments.
- Strong experience with transformer-based models and their application.
- Skilled in building token classification or sequence classification models using HuggingFace Transformers, FastText, or spaCy pipelines.
- Experience with real-time NLP — low latency, high-throughput text processing.
- Proficiency in model optimization techniques: quantization (QAT/PTQ), pruning, distillation, ONNX/TFLite/CoreML exports.
Qualifications
- Bachelor’s or Master’s degree in Computer Science, Artificial Intelligence, Computational Linguistics, or a related technical field.
- Specialization or thesis work in NLP, transformers, or text classification is a big plus.
- Strong project portfolio or GitHub showcasing NLP models built for chat, fraud detection, moderation, or spam prevention.
Experience
- 2–6 years of experience building NLP pipelines for chatbots, spam filters, trust & safety tools, or fraud classification systems.
- Experience in deploying NLP models into production environments.
- Experience building, deploying, and optimizing mobile-first ML models or inference engines for real-time applications.
Bonus if You Have
- Built models to detect social engineering, phishing, or manipulative microtext.
- Familiarity with multilingual tokenization and fraud slang handling (Hinglish, Tamil-English code mix).
- Experience with CoreML/NNAPI, mobile SDKs, or Android/iOS ML inference stack.
- Participation in Kaggle NLP competitions or published NLP research in fraud/spam.