AWS AI/Gen AI Engineer

Company Description

Webpuppies Digital, established in 2000, has been at the forefront of delivering digital transformation solutions to clients across 21 countries, including Singapore government agencies, global enterprises, and SMEs. With a strategy-first and results-driven approach, we focus on understanding business goals before crafting and executing tailored digital strategies. Our expertise spans AI consultancy, smart systems integration, web and mobile app development, cloud solutions, and emerging technologies such as VR/AR and FinTech. We take pride in being strategic digital partners, supporting businesses from planning to execution and ongoing innovation.

Role Summary

You will lead the design and delivery of embedded AI and GenAI capabilities that sit on top of the AWS data platform — covering use case shaping, model selection, prompt and retrieval design, fine-tuning where appropriate, evaluation, and production deployment with the right guardrails.

Key Responsibilities

• Use case shaping: partner with client stakeholders to translate business problems into GenAI / ML solutions with measurable outcomes.
• Bedrock-based GenAI: design RAG architectures using Bedrock Knowledge Bases, Bedrock Agents, OpenSearch / pgvector / Aurora vector stores, and Bedrock Guardrails.
• Model selection & evaluation: evaluate Anthropic Claude, Amazon Nova, Llama, Mistral and other Bedrock-hosted models against task-specific benchmarks; run offline and human-in-the-loop evaluation.
• SageMaker ML: build classical ML and deep learning pipelines using SageMaker Studio, Training Jobs, Pipelines, Model Registry, and real-time/batch inference endpoints.
• Fine-tuning & customization: where appropriate, perform fine-tuning, continued pre-training, or distillation on Bedrock or SageMaker JumpStart.
• Productionization: deploy with proper IAM scoping, VPC isolation, private endpoints, observability (CloudWatch, Bedrock model invocation logs), and cost controls.
• Responsible AI: implement guardrails, PII redaction, prompt-injection defenses, evaluation harnesses, and bias / safety reviews aligned to client policy.
• Integration: expose AI capabilities to applications and BI consumers via API Gateway, Lambda, AppSync, or embedded inside QuickSight (e.g., QuickSight Q / generative BI).
• Collaboration: work with Data Engineers on feature pipelines and with Governance Engineers on lineage, access control, and audit trails.

Must-Have Skills & Experience

• 5+ years in ML / AI engineering, with at least 2 years building production solutions on AWS.
• Hands-on experience with Amazon Bedrock — Knowledge Bases, Agents, Guardrails, model invocation, and orchestration.
• Hands-on experience with Amazon SageMaker — Studio, Training, Pipelines, Model Registry, endpoints.
• Strong Python; proficiency with LangChain, LlamaIndex, or equivalent orchestration frameworks.
• Solid grounding in RAG architectures, embeddings, vector search, evaluation methods, and prompt engineering.
• Experience with MLOps practices: CI/CD for models, monitoring, drift detection, retraining.
• Comfortable working in an RFP / pre-sales context — able to author solution narratives and defend designs.

Nice-to-Have

• AWS Certified Machine Learning – Specialty or AI Practitioner.
• Experience with multi-agent orchestration, tool use, and structured output patterns.
• Exposure to QuickSight Q / generative BI and embedded analytics use cases.
• Background in NLP, document understanding, or conversational AI.
• Familiarity with Lake Formation and governed access for AI training data.

Engagement Details:

Engagement type: Project-based / contract

Location: Singapore / hybrid – onsite as required by client cadence

Duration: Initial 6–12 months, with extension subject to RFP award and delivery milestones

Reporting to: Engagement Lead / Delivery Manager

Why Join This Engagement

• Anchor role on a flagship AWS analytics and GenAI program with a marquee client.
• Work shoulder-to-shoulder with AWS specialists across Bedrock, SageMaker, Glue, Redshift, Lake Formation, and QuickSight.
• Tight, aligned delivery pod — no dilution across competing SI narratives, clearer ownership, faster decisions.
• Real production scope — the work moves from RFP response straight into build and operate.
• Direct exposure to senior client stakeholders and AWS partner teams.