Senior Python Engineer (Latin America Candidates Only) — AI Voice & Agentic Systems (Remote)

Senior Python Engineer (LATAM) — AI Voice & Agentic Systems (Remote) About Aptitude AI / DAS Labs We build production-grade AI agents (voice + chat) that run in the real world—powering executive assistants (“Vicky”), ops/task automations (“Tim”), and bespoke client deployments. Expect realtime pipelines, measurable outcomes (latency/containment/CSAT), and security-minded engineering. The role Own backend Python services that power realtime AI voice and agent experiences. You’ll design APIs and event streams, integrate ASR/TTS/LLM providers, implement tool-calling with Model Context Protocol (MCP), and ship reliable systems used by paying customers. What you’ll do • Build & maintain Python services (FastAPI/asyncio) for low-latency voice/chat agents. • Implement LLM tool-calling end-to-end: JSON-schema tools/functions, structured I/O, retries, timeouts, idempotency, safety. • MCP: build/extend MCP servers & clients to expose tools (databases, CRMs, webhooks, n8n), with per-tenant auth and capability scoping. • Integrate voice & speech: Twilio/WebRTC; streaming ASR/TTS (Deepgram/Whisper/Google/ElevenLabs). • Orchestrate LLMs (OpenAI/Anthropic) with guardrails/evals and RAG (Pinecone/Weaviate/Qdrant). • Design evented pipelines (WebSockets, webhooks, queues) and ship to AWS/GCP with Docker & CI/CD. • Instrument everything (logs/traces/metrics via OpenTelemetry), run A/B tests, keep latency/error budgets green. • Collaborate with product to scope MVPs, iterate fast, and harden to stable releases. • Occasional ML/CV work (bonus): OCR/vision pipelines, model inference services, and data/labeling workflows. What you bring • 4–8+ years of production Python (FastAPI/Starlette or Flask + asyncio), strong testing (pytest) and software design. • Proven experience shipping commercial AI agents (voice and/or chat) used by real customers. • Hands-on with MCP (building tools/servers or integrating clients) and LLM function/tool calling in production. • Realtime stack familiarity: Twilio/WebRTC, streaming ASR/TTS; plus orchestration patterns (tools, evals, retries). • Datastores: PostgreSQL, Redis; plus a vector DB (Pinecone/Weaviate/Qdrant). • Cloud/DevOps: Docker, CI/CD, basic IaC; comfortable with tracing and performance tuning. • Communication: clear written/spoken English (B2+), collaborative, low-ego. Nice to have / Bonus points • Machine Learning & Computer Vision: PyTorch/TensorFlow, OpenCV, OCR (Tesseract/Cloud Vision), model serving (ONNX/TensorRT/Triton), on-device/edge (Core ML/TFLite), realtime video/image pipelines, MLflow/W&B. • Vapi experience for voice agents (not required). • Queues (Kafka/SQS/Pub/Sub), Kubernetes, Firebase/Firestore, or n8n. • Spanish or Portuguese; familiarity with LATAM markets and multilingual agents. • PCI-aware voice flows (DTMF redaction, compliant handoffs). Why Aptitude AI • Real impact: your code powers live agents answering real customers. • Learning velocity: fast iterations, modern stack, diverse client problems. • Product + services: shape Vicky/Tim while delivering high-visibility client wins. • Remote-first across LATAM with USD-denominated compensation. • Location & hours: Remote in Latin America with ≥4 hours overlap to 9am–5pm Pacific. • Compensation: Competitive (USD). Contract-to-hire or full-time. Our hiring process 1. Pre-interview questionnaire (10–15 min): agent architecture, MCP/tool-calling scenarios. 1. English screening test: verifies B2+ proficiency (written & spoken). 1. Technical interview: Python/asyncio + system design for realtime agents. 1. (Optional) Practical exercise or code walk-through from past work. How to apply Apply on LinkedIn (“Easy Apply”) and include: 1. GitHub/portfolio, 1. a short note on your most significant production agent (what it did, tools/MCP usage, scale, key metrics), 1. availability & preferred arrangement (contract/full-time).