About this opportunity
Sparktek lists this Responsibilities opportunity in san jose, California. Review the employer’s description below for duties, qualifications and application requirements.
Job description
Machine Learning Engineer
We are seeking a highly skilled Machine Learning Engineer to design and build a low-latency query understanding and intelligent routing system that operates without reliance on large language models. The role focuses on extracting intent, entities, application context, routing decisions, and supporting evidence from user queries in real time.
This is a full lifecycle role spanning data modeling, ML development, optimization, local deployment, and MLOps. The ideal candidate will have strong experience in applied NLP, lightweight model architectures, and production-grade ML systems, with a focus on sub-second inference, CPU-based execution, and scalable domain evolution.
Responsibilities
Design and implement a query understanding pipeline to extract intent, routing decisions, entities, application mapping, and historical evidence from user queries and conversations.
Define and build the training data model and annotation schema for structured outputs (intent, routing, entities, applications, evidence).
Lead data collection, synthesis, analysis, and cleaning to develop high-quality datasets for model training and evaluation.
Develop and evaluate baseline and advanced non-LLM models for:
Intent classification
Query routing
Entity extraction
Application detection
Evidence retrieval
Build and maintain train, test, and evaluation pipelines with strong focus on:
Accuracy and F1 score
Confidence scoring and calibration
Latency and throughput
Optimize models to meet strict constraints:
Sub-second inference latency
CPU-only execution
Compact model size ( Deploy models locally within the application codebase, ensuring seamless integration without reliance on hosted AI services.
Design and implement a Level 4 MLOps framework, including:
Monitoring and alerting
Drift detection
Retraining pipelines
Data feedback loops
Develop strategies to handle domain evolution, including:
New agents / skills
New entity types
Updates to domain definitions
Leverage historical queries and routing decisions to improve prediction accuracy and evidence generation.
Collaborate with product, engineering, and domain teams to translate business workflows into scalable ML solutions.
Deliver a working demo/prototype baseline, and iteratively mature it into a production-ready system.
Required Skills
Strong expertise in Machine Learning and Applied NLP, especially in:
Text classification
Intent detection
Query routing
Entity extraction
Semantic similarity and retrieval
Proven experience with non-LLM approaches, including:
Encoder-based models
Embedding-based pipelines
Classical ML (e.g., XGBoost, Logistic Regression)
Lightweight deep learning models
Experience designing training datasets, labeling frameworks, and structured output schemas for multi-task NLP systems.
Strong understanding of data preprocessing and quality improvement, including:
Normalization
Deduplication
Class imbalance handling
Synthetic data generation
Experience building robust evaluation frameworks, including:
Precision, Recall, F1
Confidence scoring
Ranking quality
Latency measurement
Hands-on experience with entity extraction for structured enterprise domains, such as:
Device identifiers (PID, Serial Number, MAC, Hostname)
Smart / Virtual accounts
Orders, contracts, subscriptions
Product families and licenses
Experience handling multi-label and hierarchical classification problems.
Strong ability to build low-latency, CPU-optimized inference systems with strict memory and performance constraints.
Experience deploying ML models locally or on-prem within application codebases (not limited to cloud-hosted inference).
Solid understanding of MLOps practices, including:
Monitoring and observability
Drift detection
Retraining pipelines
Model lifecycle management
Strong programming skills in Python, with hands-on experience in ML/NLP frameworks and pipeline orchestration.
Ability to adapt systems to continuous domain changes, including new skills, applications, and entities.
Prior experience in enterprise support systems, operational routing, licensing platforms, or device/account management domains is highly preferred.
Worksite address
san jose, CA, 95199, US
Who can apply
Review the original listing for work authorization, qualifications and employer requirements.