AI / LLM / Software Engineer

The AI / LLM / Software Engineer engineers secure, governed, evaluated and maintainable AI applications using large language models, retrieval, agents, computer vision and model-serving technologies. The role turns AI capability into supported production services rather than isolated prototypes.

Language, agent and knowledge applications

• Build retrieval-augmented generation systems, knowledge assistants, AI agents, and classification and generation workflows.
• Build applications based on user needs and the available AI context, in partnership with BI & AI Context Engineers.
• Develop supporting application services, interfaces and integrations.

Computer vision and video analytics

• Build and operate computer-vision services — queue length and wait time, speed of service, occupancy and footfall, dwell time, and customer-journey analytics for store operations.
• Build vision applications for manufacturing — defect and surface inspection, process and line monitoring, PPE and safety compliance, and material or asset presence detection.
• Build display, planogram, shelf and inventory-presence detection where the business case is established.
• Own the video pipeline end to end: camera and NVR integration, frame sampling and extraction, annotation and labelling workflow, training and fine-tuning, inference and result persistence into the Lakehouse.
• Establish vision-specific evaluation — precision and recall by condition, lighting and camera-angle robustness, false-positive cost, and drift as stores or lines change.

Sensing-device and edge AI

• Build AI over IoT and other sensing devices — temperature, weight, vibration, energy, RFID and scanner data — including anomaly detection and event recognition.
• Deploy and operate models at the edge where bandwidth, latency, connectivity or data-retention constraints prevent central inference.
• Manage edge device fleets, model versioning and rollback, remote monitoring and offline degradation behaviour.
• Work with Systems Integration Engineers to persist only the events and features the business needs rather than raw streams.

Evaluation, safety, privacy and control

• Implement evaluation suites, guardrails, access control, observability, cost management and human-approval workflows.
• Test for accuracy, harmful output, prompt injection, data leakage and regression before release.
• Enforce policy on personal, biometric and video data — consent and notice, purpose limitation, masking or blurring, retention limits, restricted access and audit logging — in coordination with the Data Governance Specialist, Legal and Cybersecurity.
• Ensure video and sensing applications are used for the approved operational purpose and are not repurposed for individual surveillance without an approved basis.

Productionization

• Productionize model endpoints, vector indexes, vision services, application services, CI/CD, monitoring, fallback and incident procedures.
• Define and meet latency, availability, throughput and cost targets, including per-camera and per-device inference cost.
• Maintain support runbooks and participate in incident response.

Reuse and research

• Reuse enterprise patterns and components rather than creating isolated prototypes.
• Research and push the capabilities of AI further, and bring proven advances into the enterprise pattern library.

• Production AI applications across language, vision and sensing use cases, plus reusable components.
• Computer-vision services with camera and NVR integration, annotated datasets and labelling workflow.
• Edge deployment packages, device fleet management and rollback procedures.
• Evaluation suites and published evaluation results, including vision precision and recall by operating condition.
• Model endpoints, vector indexes and serving infrastructure.
• Observability dashboards covering quality, latency, usage and cost, including per-camera and per-device cost.
• Privacy and retention controls for video, biometric and sensor data, with audit evidence.
• Operating runbooks and incident procedures.

• Reliability, latency and availability of AI, vision and edge services.
• Accuracy of vision and sensing detections in real operating conditions, not just on test sets.
• Safety, security and policy compliance of AI outputs.
• Privacy compliance for video, biometric and personal data, including retention and access discipline.
• Cost efficiency of AI workloads, including video storage and inference cost.
• Maintainability and reuse rather than one-off builds.

• Internal: Data Scientists; Data Engineers; Systems Integration Engineers; BI & AI Context Engineers; Data Governance Specialist; Cybersecurity; IT Infrastructure; Physical Security and Facilities for camera estates; store operations and plant Safety, Quality and Production teams; Legal and the Data Protection Officer; Data & AI Translators; Data Science & AI Capability Head.
• External: AI platform and model vendors; camera, NVR, edge hardware and IoT device vendors.

Education

Bachelor's degree in STEM (Math, Physics), Computer Science, Software Engineering, Electronics or Computer Engineering, Information Technology or a related field.

Experience

Three or more years in software or machine-learning engineering, including at least one year building and operating LLM, agent, retrieval, computer-vision or sensor-AI applications in production. Experience with CCTV, NVR or edge video pipelines is a strong advantage for store-operations and plant-assigned positions. Strong software engineering fundamentals are required.

Certifications

Preferred: Databricks Machine Learning or Generative AI certification, cloud platform associate certification. Optional: computer-vision or edge AI specialization, security or MLOps certification.

• Python and strong software engineering practice; at least one additional language is an advantage.
• LLM application patterns — prompting, retrieval-augmented generation, agents, tool use, structured output.
• Vector stores, embeddings and retrieval evaluation.
• Computer vision — object detection, tracking, segmentation, classification, OCR; annotation and labelling workflow; model fine-tuning.
• Video engineering — RTSP and camera streams, NVR integration, frame sampling, codecs, storage and retention design.
• Edge inference and device fleet management, including offline behaviour and model rollback.
• IoT and sensor data handling, anomaly detection and event recognition.
• Model serving, APIs, containerization and CI/CD.
• Observability, evaluation frameworks, guardrails and cost monitoring.
• Privacy-preserving techniques for video and personal data — masking, blurring, on-device processing, minimal retention.
• Databricks, MLflow and cloud services.

• Engineering discipline applied to a fast-moving field.
• Security and privacy awareness by default, especially where cameras and people are involved.
• Evidence-driven — evaluates in real operating conditions rather than asserting.
• Pragmatism about which problems actually need AI, and which need a sensor or a process change.
• Comfort working on the shop floor and in stores to see how the system behaves in practice.
• Collaboration and reuse over individual invention.

Levels below are indicative and subject to OD and HR job evaluation.

Level

Expectation

AI / LLM / Software Engineer

Builds and supports assigned AI, vision or sensing applications under technical direction. Three or more years of relevant experience.

Senior AI / LLM / Software Engineer

Owns an AI product area such as store vision analytics or plant inspection, sets evaluation and safety patterns, mentors engineers. Six or more years of relevant experience.

Lead AI Engineer

Sets enterprise AI application architecture and standards across language, vision and edge, and deputizes for the Capability Head. Nine or more years of relevant experience.

"Motolite offers you not just a job, but a career with boundless opportunities"