AI Systems Engineer Job

Pickle is hiring an AI Systems Engineer to build the core systems behind our real-time memory layer.

This role focuses on turning raw user data (emails, messages, notes, conversations, and other context streams) into structured, long-term memory that AI can reliably use in live interactions. You will work across memory pipelines, retrieval systems, multi-provider LLM infrastructure, real-time backends, and privacy-first security architecture.

This is a hands-on, production-focused role. You will design and ship systems that must be low-latency, scalable, and reliable, while handling highly sensitive personal data with strong security and compliance guarantees.

Key Responsibilities
1. Build production-grade pipelines that convert user data sources (email, messages, notes) into structured long-term memory, including episodic chunking, fact extraction, and embedding generation.

2. Design and optimize retrieval/search systems using PostgreSQL (pgvector) and OpenSearch to support low-latency, reliable memory access in real time.

3. Develop multi-provider LLM infrastructure (OpenAI/Anthropic/AWS Bedrock/Azure OpenAI), including routing, rate limiting, failover, and token usage/cost tracking.

4. Build and operate real-time backend services (SSE/streaming) and “memory search tools” that can be invoked dynamically within AI interactions.

5. Deploy and maintain AWS production infrastructure (EKS/Kubernetes), including PostgreSQL/OpenSearch operations and Docker/ECR pipelines.

6. Own security and compliance deliverables for sensitive personal data, including encryption controls and support for third-party verification/audits (e.g., Google OAuth/CASA Tier 2 or similar).

7. Prepare technical materials and support partner-facing engineering discussions with AI/cloud/hardware partners (e.g., AWS, OpenAI, Qualcomm) for integration and quota coordination

Minimum Requirements (Skills-Based; No Degree Required)

1. Backend Development: Experience building backend services in Python and/or TypeScript.

2. AI Memory Pipeline: Experience building pipelines that convert raw user text data (e.g., emails/messages/chat logs/notes) into structured memory, including episodic chunking, fact extraction into structured nodes, and embedding generation.

3. Retrieval/Search Systems: Experience implementing and operating retrieval using PostgreSQL with pgvector and OpenSearch (index design and query usage/optimization).

4. Multi-Provider LLM Infrastructure: Experience integrating multiple LLM providers (e.g., OpenAI/Anthropic/AWS Bedrock/Azure OpenAI) with routing/failover, per-provider rate limiting, and token usage/cost tracking (e.g., using Redis or equivalent).

5. Real-Time Backend: Experience implementing SSE-based real-time chat or streaming responses.

6. AWS Infrastructure: Experience deploying/managing production systems on AWS, including EKS/Kubernetes, operating PostgreSQL and OpenSearch, and managing Docker/ECR pipelines.

7. Security/Compliance: Demonstrated experience leading security documentation and verification/audit work for sensitive data integrations (e.g., Google OAuth verification and/or CASA Tier 2 or comparable third-party security assessment), including implementing encryption controls.

Note: “Experience” refers to the project duration of at least three (3) months.

Health/Dental Benefits, Retirement Benefits, Stock or Stock Options, Paid Holidays, Vacations, and Sick Leave

Pickle, Inc. is building a consumer-first AI memory system that turns a user’s digital and real-world context into persistent, actionable intelligence. At the core of Pickle is Pickle OS, a real-time memory layer that continuously ingests and organizes personal context across email, chat, notes, conversations, and interactions, so AI can understand who the user is, what they care about, and what they intend to do next. Since our public launch in early 2026, Pickle has reached thousands of active users and meaningful early revenue, with especially strong retention among power users who connect multiple data sources. We are now scaling the core memory infrastructure that powers Pickle OS, Pickle Chat, and upcoming consumer-facing AI experiences.