Conversational AI & Chatbots:

• Lead the design and delivery of Snoonu's multi-agent AI chatbot service for driver support across real-time chat channels.
• Own the four-agent architecture — Coordinator, Data Collector, Rules Agent, and Action Executor — running on AWS Bedrock Agents or similar architecture.
• Ensure chatbot flows handle driver intents reliably: order removal, vehicle mismatch, ETA extensions, merchant disputes, and escalations.
• Drive LLM evaluation cycles, prompt strategy, and Bedrock Guardrail design to ensure responses are consistent, safe, and operationally correct.
• Integrate chatbot services with order management, CRM (Salesforce), Slack operations channels, and driver-facing apps via secure, event-driven patterns.

Agentic AI & SOP Automation:

• Lead the buildout and operation of the SOPs-as-Code framework — encoding operational SOPs as machine-readable policies executed by the multi-agent system.
• Own the Config File architecture and Config Reader Agent pipeline that converts PDFbased SOPs into deployable agent configurations on AWS Bedrock.
• Govern the structured rules engine (condition/operator/value schema) to ensure deterministic, auditable decisions with no LLM interpretation ambiguity.
• Design and enforce human-in-the-loop checkpoints, escalation triggers, confidence thresholds, and operator override capabilities.
• Establish versioning, rollback, and safe deployment practices for SOP configuration changes in production.

AWS Infrastructure & MLOps:

• Own the cloud backbone for AI services: Lambda, ECS/Fargate, SQS/SNS, DynamoDB, MongoDB, S3, CloudWatch, and AWS Bedrock.
• Build and maintain CI/CD pipelines for prompt versioning, agent configuration rollout, and automated eval gates before production deployment.
• Define observability standards — per-agent-turn latency SLAs, Bedrock cost tracking, drift detection, and failure alerting.
• Lead capacity and cost planning as interaction volumes scale across driver and operations channels.

R&D & Innovation:

• Evaluate frontier LLMs (Claude Sonnet/Opus, open-weight models) and orchestration frameworks (Bedrock Agents, LangGraph, CrewAI) against Snoonu's operational constraints.
• Identify and prototype the next AI capability Snoonu should dominate — from experiment to validated proof-of-concept with clear go/no-go criteria.
• Produce architecture decision records, prompt engineering playbooks, and technical documentation for team-wide use.

Education

• Bachelor's or Master's degree in Computer Science, AI, Software Engineering, or a related field.

Experience

• 6–10 years of software engineering experience, with at least 3 years in a tech lead or engineering management role.
• Demonstrated track record of shipping production conversational AI, IVR, or agentic systems end-to-end — not just models, but the full stack from architecture to monitoring.
• Hands-on experience with multi-agent orchestration on AWS Bedrock or equivalent agentic frameworks (LangGraph, CrewAI, AutoGen).
• Prior experience leading a team of 3–8 engineers, with a coaching-first approach to technical growth.
• Strong Python and backend development skills; able to write, review, and hold the bar on production-grade code.
• Engineering multiplier: makes every engineer on the team faster and better through design guidance, code reviews, and clear technical direction.
• Ownership without ego: defines problems, architects solutions, ships results — accountable for outcomes across the full platform, not just assigned tickets.
• Operational intelligence: understands the business context of driver support, logistics operations, and the real cost of failure in real-time systems.
• Senior communicator: articulates trade-offs (agent reliability vs. automation rate, cost vs. latency) clearly to non-engineers and influences roadmap decisions through clarity of thought.
• Structured under ambiguity: brings process and rigor to fast-moving R&D environments where requirements evolve alongside the build.
• Bias for action: prototypes fast, validates early, and ships iteratively — while maintaining the quality bars that prevent production incidents.
• Collaborative and direct: raises architectural risks early, pushes back on under-specified requirements, and surfaces trade-offs before they become delivery blockers.

Skills

Conversational AI & LLMs (Core)

• Deep experience with AWS Bedrock — Agents, Knowledge Bases, Guardrails, and model invocation; strong preference for Claude Sonnet/Opus via Bedrock.
• Multi-agent system design: orchestration patterns, agents-as-tools, inter-agent handoffs, context propagation, and failure isolation.
• Prompt engineering: system prompt design, structured output, tool-use, multi-turn reasoning, eval-driven iteration, and red-teaming for safety.
• Structured rules engines: condition/operator/value schemas for deterministic, noninterpretive decision logic — mandatory in production agentic systems.
• RAG pipelines: embedding models, vector databases (OpenSearch, Pinecone, pgvector), hybrid retrieval, and knowledge base tuning.

IVR & Voice

• IVR architecture: call flow design, intent/slot management, DTMF handling, escalationto-human routing, and SLA monitoring.
• Experience with STT/TTS pipelines or voice bot platforms is a strong plus.

AWS Services (Core)

• Lambda, API Gateway, SQS/SNS, Step Functions, DynamoDB, S3, CloudWatch — event-driven and serverless architectures.
• AWS Bedrock Agents: agent creation, alias management, tool action group configuration, and Guardrail policy management.
• IAM, VPC, Secrets Manager — security and environment best practices for AI service deployments.
• CI/CD for AI services: prompt versioning, agent config deployment pipelines, automated evals, and rollback gates.

Backend & Engineering

• Python — async, OOP, clean code; REST API design with FastAPI or Flask.
• MongoDB and DynamoDB — schema design, indexing, querying, and operational monitoring.
• Docker, ECS/Fargate; Git, automated testing, and CI/CD workflows.
• Salesforce integration (APIs, events, data sync) is a strong plus.

Bonus Points if You Have:

• Knowledge of logistics, food delivery, or real-time operations domains.
• Arabic language NLP or experience building for Arabic-speaking markets.
• Familiarity with open-weight model inference (Ollama, vLLM, HuggingFace Transformers).
• Exposure to multi-modal AI or voice-first interaction design.