AI Backend Engineer

AI Backend Engineer — Conversational Retail Sales Agent Platform We're building an AI-powered autonomous sales agent. We're an early-stage, patent-pending startup with a working concept and a clear roadmap, and we're hiring our first backend engineer to help build the core platform. What you'll do: - Architect and build the cloud backend on AWS (serverless and IoT services, real-time messaging, data pipelines) - Integrate conversational AI and natural-language capabilities into a responsive customer-facing experience - Build and maintain RESTful APIs connecting our web/mobile front end, internal CRM, and third-party retail systems - Design for reliability, low latency, and scale across multiple store locations You should have: - Strong Python and production AWS experience (IoT Core, Lambda, API Gateway, DynamoDB or similar) - Experience integrating AI/LLM and NLP services into real products - Solid RESTful API design and backend architecture skills - Comfort working independently in an early-stage environment and shipping iteratively *Plus to have: - Infrastructure-as-code (Terraform), monitoring/observability - Experience with real-time/event-driven systems or IoT device integration *Engagement: Please share relevant backend and AWS work. Candidates will sign an NDA before receiving detailed product specs. 2.1 Required Technical Skills Python — Expert Level (Non-Negotiable) * Python 3.10+ with deep understanding of async/await patterns and asyncio * FastAPI framework for building high-performance REST APIs and WebSocket endpoints * Pydantic v2 for data validation and serialization across all API layers * SQLAlchemy 2.0 ORM with async session management for PostgreSQL * pytest and pytest-asyncio for comprehensive unit and integration testing * Poetry or pip-tools for dependency management and reproducible builds AI & Large Language Model Integration * OpenAI API (GPT-4o, GPT-4-turbo) — function calling, streaming, token management * Anthropic Claude API — Messages API, system prompts, tool use implementation * LangChain or LlamaIndex for AI conversation flow orchestration and memory management * Vapi.ai or Retell.ai SDK integration for real-time voice conversation pipeline * ElevenLabs API for voice cloning, text-to-speech synthesis, and real-time streaming * Prompt engineering — system prompt design, few-shot examples, chain-of-thought * Retrieval Augmented Generation (RAG) for product database integration into AI context * Vector database experience — Pinecone, Weaviate, or pgvector for product embeddings * AI conversation state management — multi-turn context, session continuity across devices * Speech-to-text integration — Deepgram, Whisper, or equivalent for voice input processing AWS Cloud Infrastructure * AWS Lambda — Python runtime, function composition, cold start optimization * AWS API Gateway — REST and WebSocket API configuration, CORS, throttling, authorizers * AWS IoT Core — MQTT broker setup, device registry, thing types, policy configuration * AWS RDS PostgreSQL — multi-AZ setup, parameter groups, read replicas, backup config * AWS S3 — bucket policies, lifecycle rules, presigned URLs, CloudFront integration * AWS Cognito — user pools, identity pools, JWT token validation, custom authorizers BIADSAY Hiring Specifications — CONFIDENTIAL — Page 3 of 26 * AWS CloudFront — CDN configuration, cache behaviors, Lambda@Edge * AWS SQS/SNS — message queuing for async job processing and event notifications * AWS CloudWatch — logging, metrics, alarms, dashboards, X-Ray tracing * AWS Secrets Manager — secure credential storage and rotation * AWS CDK or Terraform for infrastructure as code — repeatable environment deployment MQTT & IoT Protocol * MQTT protocol — publish/subscribe patterns, QoS levels 0/1/2, retained messages * AWS IoT Core MQTT topic design — hierarchical topic structure, wildcard subscriptions * IoT Core device shadow — for RFID reader state management and offline handling * MQTT over WebSocket for browser-based real-time communication * Understanding of UHF RFID EPC Gen2 protocol for tag activation command construction * WiFi-connected RFID reader API integration — REST or MQTT command interfaces Database & Data Architecture * PostgreSQL 15+ — complex queries, indexing strategy, EXPLAIN ANALYZE optimization * Row-level security (RLS) for multi-tenant data isolation between store accounts * Database schema design for CRM — customers, jobs, transcripts, samples, appointments * TimescaleDB or partitioning for time-series transcript and event log storage * Redis for session caching, real-time pub/sub, and rate limiting * Database migration management with Alembic Real-Time Communication * WebSocket server implementation with AWS API Gateway WebSocket API * Connection management — heartbeat, reconnection, multi-device session routing * Real-time event broadcasting for kiosk-to-phone avatar migration sequence * Socket.io familiarity as an alternative real-time layer SMS & Communications * Twilio API — SMS sending, receiving, webhook processing, conversation threads * Twilio Conversations API for structured multi-turn SMS dialogue management * SMS webhook handler — inbound message routing to AI conversation engine * vCard generation and SMS delivery for business card handoff feature 2.2 Nice-to-Have Skills * Face synthesis API experience — D-ID, HeyGen, or equivalent avatar generation * ElevenLabs voice cloning pipeline — recording processing, model training, deployment BIADSAY Hiring Specifications — CONFIDENTIAL — Page 4 of 26 * UHF RFID hardware experience — reader configuration, antenna placement, read rates * Stripe or payment processing API integration * Docker and container orchestration for local development environments * GraphQL API design as an alternative to REST 2.3 Responsibilities * Design and build the BIADSAY AI conversation engine — voice input to product match to RFID activation in under 300ms end-to-end * Build and maintain all AWS Lambda functions, API Gateway routes, and IoT Core configurations * Implement the MQTT command pipeline from AI engine to RFID reader to tag activation * Build the NEXSA CRM data model and all automatic data capture pipelines from AI conversations * Integrate ElevenLabs voice cloning and D-ID or HeyGen face synthesis for avatar generation * Build the Twilio SMS follow-up automation engine with AI-driven response handling * Implement multi-tenant row-level security across all database tables * Build the AI-generated session summary feature using LLM summarization of transcripts * Write comprehensive unit tests achieving minimum 80% code coverage * Document all APIs using OpenAPI 3.0 specification * Participate in weekly technical reviews with founder and project manager 2.4 Qualifications * Minimum 4 years professional Python backend development experience * Minimum 2 years production AWS deployment experience — not just tutorials * At least 1 completed AI/LLM production project — portfolio or GitHub required * Demonstrable IoT or MQTT project experience — describe a real deployment * Bachelor's degree in Computer Science, Software Engineering, or equivalent experience * Must be able to provide 2-3 professional references from prior AI or cloud projects * Available for 30-40 hours per week minimum for 6-month engagement * Must sign NDA and IP Assignment Agreement before project kickoff