Sr AI Engineer - Memory Systems (LangGraph, Python, TTS/STT AWS, Postgres/Redis) - FT - WW

About the Role

We’re looking for a senior AI Engineer to support and enhance a production-grade voice AI product. This role is heavily focused on post-production support, testing, and deep expertise in Speech-to-Text (STT) and Text-to-Speech (TTS) systems.

You will work directly on a live system that has already been deployed to production, ensuring reliability, alignment with requirements, and continuous improvement. This is a full-time contract role with the possibility of long-term engagement.

Must-haves

• 5+ yrs Python; shipped prod systems using Postgres JSONB and Redis (TTL, sizing, cache coherency).
• Production experience with LangGraph (or equivalent orchestration) with state/checkpoint patterns, TTS/STT and function along with tool calling.
• AWS (ECS/Lambda/Step Functions/Bedrock), containerized services, IaC, CI.
• Strong schema design & LLM data contracts (Pydantic/JSON Schema); observability (OpenTelemetry/structured logs).
• Pragmatic safety/privacy: redaction pipelines, auditable changes, least-privilege access.

Nice-to-haves

• Django + DRF (or FastAPI), Celery/Step Functions schedulers, pgvector.
• LiveKit/real-time systems, TTS/STT pipelines.
• COPPA/GDPR-K patterns.

Outcomes (first 90 days)

• Ship memory v1: default prompt ≤ ~800 tokens from Child Context; redis retrieval P95 <100 ms, join-time load <500 ms.
• Nightly/weekly/monthly rollups with idempotent backfill and a clean retrieval ladder in the agent.
• Dashboards for freshness, token budgets, retrieval durations, and cache hit rates.