AI / Data Engineer, AI/ML-powered IoT Platform

Who We Are

JIG-SAW operates 24/7 Operations Centers in Japan and Canada that proactively monitor systems, issue alerts, and deliver live incident response—keeping web services and IoT environments secure and running smoothly.

• Corporate site: https://jig-saw.com/en/
• NEQTO.ai: https://neqto.ai/

About the Role

We are looking for a dual-threat AI and Data Engineer to build a game-changing, AI-driven IoT platform. You won’t just be building chatbots; you will be building the "nervous system" of industrial environments.

The core of this role is turning raw, often messy device signals into actionable intelligence. You will own the end-to-end pipeline: from integrating industrial protocols and normalizing high-velocity data streams to deploying RAG-based insights and predictive ML models. If you enjoy the challenge of making intelligence observable, reliable, and scalable across factories, buildings, and logistics hubs, this role is for you.

Key Responsibilities

• Data Pipeline Architecture: Design and implement robust data ingestion and processing pipelines that receive, normalize, and transform raw data from diverse IoT sources into structured formats for AI consumption.
• Protocol Integration: Implement and manage connectivity for industrial and wireless protocols including Modbus, BACnet, LoRaWAN, and MQTT.
• AI/ML Development: Ship insight features combining LLM/RAG with classical ML for anomaly detection, forecasting, and predictive maintenance.
• Data Engineering at Scale: Handle "big data" challenges—managing time-series datasets, stream processing, and ensuring data integrity for downstream models.
• Retrieval & Orchestration: Implement robust retrieval pipelines (chunking, embeddings, reranking) and agentic workflows to turn business questions into measurable outcomes.
• Production Excellence: Own the loop end-to-end—training, deploying, and monitoring models in production with clear KPIs and guardrails.

Required Skills & Qualifications

• Data Engineering & Processing: Expert-level experience in ETL/ELT, data normalization, and processing high-throughput streaming data.
• LLMs & RAG: Deep understanding of LangChain/LlamaIndex, agent frameworks, and tool orchestration for generating insights beyond simple text generation.
• ML/DL Frameworks: Proven experience using TensorFlow, PyTorch, or Scikit-learn for time-series forecasting and anomaly detection.
• Cloud & Vector Infra: Proficiency in GCP/AWS/Azure and vector databases (e.g., pgvector, Pinecone, Milvus, or Weaviate).
• Education: Bachelor’s degree or higher in Computer Science or a closely related technical field.

Experience Guidelines

• 5+ years of professional Data Engineering/ML experience: building and operating production-grade platforms (ingestion, training, deployment, and monitoring).
• 5+ years of Industrial IoT (IIoT) domain experience: working with smart factories, logistics, energy management, or smart buildings.
• 3+ years delivering GenAI (LLM/RAG) features for real-world users (beyond hobbyist projects).

Nice-to-Have

• MLOps: Experience with MLflow, W&B, or SageMaker for experiment tracking and model versioning.
• Industrial Connectivity: Hands-on experience integrating with Modbus (TCP/RTU), BACnet, LoRaWAN, and MQTT.
• Advanced Analytics: Experience with Temporal Fusion Transformers (TFT), Prophet, or state-space models for time-series at scale.
• Security: Knowledge of PII handling, RBAC, and secure device onboarding.

Compensation & Benefits

• Remote Work: Fully remote position.
• Health & Wellness: Comprehensive Health, Dental, and Vision insurance.
• Time Off: Competitive PTO (Vacation, Sick Leave, and Company Holidays).
• Salary: Competitive market-based salary dependent on location and experience.

Portfolio Required: Please include links to your GitHub/GitLab, product sites, or technical case studies. Applications without a portfolio will not be considered.

Job Type: Full-time

Pay: $100,000.00 - $150,000.00 per year

Benefits:

• Dental insurance
• Health insurance
• Paid time off
• Vision insurance

Application Question(s):

• [Required] This is the first-round screening question #1. If this field is left blank, you will not be considered.

The key evaluation criterion is whether you can explain your reasoning logically. We value how you think more than what you already know. If a response is detected to have been generated using ChatGPT or an equivalent tool, it will be automatically rejected.

For the function below, propose technical approaches you would use to implement it. Assume real production use: consider performance, scalability, operating cost, and LLM context-window limits.

On an IoT data dashboard, display:

1) Insights through the previous day (e.g., findings, early signs of failure, predictive maintenance).

2) A list of alerts through the previous day (not simple threshold checks, but early failure signs and predictive maintenance).

• [Required] This is the first-round screening question #2. If this field is left blank, you will not be considered.

The key evaluation criterion is the same as #1.

For the function below, propose technical approaches you would use to implement it. Assume real production use: consider performance, scalability, operating cost, and LLM context-window limits.

Enable users to obtain on-demand insights (findings, early signs of failure, predictive maintenance) by asking a chatbot, based on all data collected to date.

• [Required] This is the first-round screening question #3. If this field is left blank, you will not be considered.

The key evaluation criterion is the same as #1.

For the function below, propose technical approaches you would use to implement it. Assume real production use: consider performance, scalability, operating cost, and LLM context-window limits.

From the data format sent by unknown IoT devices, auto-register/provision devices.

1) Automatically detect and register the device with its metadata when multiple sensor values compressed/obfuscated by encoding them as binary/hex.

2) Automatically detect and register the device with its metadata when multiple sensor values are human-readable (non-binary/non-hex) payloads.

• [Required] This is the first-round screening question #4. If this field is left blank, you will not be considered.

The key evaluation criterion is the same as #1.

From the data format sent by unknown IoT devices, auto-detect/map sensor attributes.

1) Automatically detect sensor attributes (e.g., temperature), normalize them, and associate attributes with the database scheme. Assume multiple sensor values compressed/obfuscated by encoding them as binary/hex.

2) Automatically detect sensor attributes (e.g., temperature), normalize them, and associate attributes with the database scheme. Assume multiple sensor values are human-readable (non-binary/non-hex) payloads.

• [Required] This is the first-round screening question #5. If this field is left blank, you will not be considered.

The key evaluation criterion is the same as #1.

For the function below, propose technical approaches you would use to implement it. Assume real production use: consider performance, scalability, operating cost, and LLM context-window limits.

Settings recommendations: a feature that recommends configuration settings the user tends to prefer.

• [Required] Please describe in detail your professional experience using generative AI in Industrial IoT environments such as smart factories, smart buildings, logistics, and energy management. If this field is left blank, you will not be considered.
• [Required] Please describe in detail your professional experience using machine learning in Industrial IoT environments such as smart factories, smart buildings, logistics, and energy management. If this field is left blank, you will not be considered.

Education:

• Bachelor's (Required)

Experience:

• professional ML (shipped, deployed, monitored): 5 years (Required)
• professional GenAI (LLMs/RAG): 3 years (Required)
• industrial IoT (e.g., smart factories, logistics): 5 years (Required)

Work Location: Remote