Senior ML Engineer (IRC276915) Offline

Job Description

Master’s or Ph.D. in Data Science, Computer Science, Statistics, Applied Mathematics, or a related field.
10+ years of experience in designing and implementing machine learning systems in a production environment.
Profound knowledge of machine learning algorithms, including but not limited to supervised and unsupervised learning, deep learning, NLP,
GenAI, and reinforcement learning.
Experience with ML frameworks and libraries (e.g., TensorFlow, PyTorch, Scikit-learn).
Proficient in programming languages used in data science and ML, primarily Python and R.
Strong understanding of data structures, algorithms, and software engineering principles.
Experience with at least 2 cloud platforms (e.g., AWS, Azure, Google Cloud) and understanding of how to leverage their ML services.
Knowledge of containerization and orchestration technologies (e.g., Docker, Kubernetes) for deploying ML models.
Familiarity with MLOps principles and tools to streamline the ML lifecycle from development to production.
Participate in pre-sales activities, including developing ML offering materials and engaging with clients to understand their needs, presenting
tailored solutions, and demonstrating the potential impact of our ML technologies.
Excellent communication and leadership skills, with the ability to work in a fast-paced, collaborative environment.

Preferred:
Certifications in cloud technologies and machine learning.
Experience with big data technologies (e.g., Hadoop, Spark).
Published work in relevant fields.

Job Responsibilities

Responsibilities: Design and develop machine learning models for anomaly and error detection in protocol trace data. ML Engineer will handle model selection, implementation, training, testing, and optimization, ensuring models meet requirements and achieve learning standards aligning with the following criteria:

• Ability to accurately detect and classify all events within a protocol trace that fall outside the range of “normal operation” (defined through training cycles performed using curated training data covering a wide scope of transactional scenarios in which classes of significant trace data elements are identified and their values are guaranteed to fall within acceptable ranges and/or within protocol defined sequence patterns)
• Ability to accept parametrized range criteria from users pertaining to an array of trace data elements and use these parameters to adjust detection sensitivity against these elements (e.g. set maximum acceptable latency within a trace to ignore latency values which might otherwise be detected as anomalous)
• Ability to recognize causal relationships between series transactions within a trace leading to a behavioral error or anomaly, and to contextualize the event sequence for use in root cause analysis
• Ensure models are dynamically scalable to trace data set of increasing size

• Ensure models are adaptable such that the base ML can be quickly trained to digest and analyze the specificities of new protocols, while reusing and building upon learning foundations generic to protocol analysis