Job Description
This is a rare opportunity to help define the foundations of emotionally intelligent AI. Everything beyond the core LLM—memory, emotional layer, and relational engine—is built in-house. The Backend Engineer will architect the systems that make our AI feel human: building the foundational codebase for the next wave of intelligent systems—ones that feel, remember, and connect.
What You’ll Do
1. Backend & Infrastructure Ownership
Contribute to and expand our C# / .NET / ASP.NET backend API layer (or similar experience with Java/Spring and willingness to adapt).
Develop modular Python microservices (FastAPI or AWS Lambda) with AI-native architecture in mind to build our intelligence layer.
Deploy models and set up ML training pipelines.
Apply best practices such as dependency injection, Strategy Pattern, and inversion of control for maintainability.
Own the backend surface area—authentication, APIs, infrastructure, orchestration—and design features for scalability and velocity.
Build and maintain low-latency REST and GraphQL APIs consumed by our iOS client.
Architect a microservice-style ML model serving backend using Docker containers or AWS Lambda (SnapStart), backed by async eventing and pub/sub.
Oversee CI/CD, rollback strategies, logging, and error handling—owning the backend end-to-end.
2. AI & ML Systems Integration
Architect and manage vector databases (PgVector or similar) to power retrieval-augmented generation, evolving memory, and personalization.
Build tools and enhance custom memory pipelines tied to user context, embeddings, and interaction history.
Integrate and scale inference with OpenAI, Claude, Llama, and other models; manage caching, fallbacks, and prompt routing logic.
Implement emotion and sentiment tagging workflows using APIs or inline lightweight classifiers.
Maintain orchestration layers for third-party model providers (e.g., OpenAI, ElevenLabs).
3. Cloud Infrastructure, DevOps, and Data Stack
Manage AWS infrastructure and expand our current stack: Lambda, ECS, S3, RDS (Postgres), CloudFront, IAM, Route53—owning architectural decisions and trade-offs.
Utilize search databases like OpenSearch.
Implement infrastructure-as-code with Terraform and CI/CD pipelines through GitHub Actions.
Ensure observability through metrics, structured logging, tracing, and alerting (OpenTelemetry, Sentry, Grafana, CloudWatch).
Optimize latency across APIs, tune Postgres indexes, add Redis caching, and integrate pub/sub or streaming for near-instant data sync.
Secure infrastructure for SOC-2 readiness—access controls, data lifecycle policies, and encrypted storage.
4. Personalization & Emotional Intelligence Layer
Design and implement emotion-aware backend systems that update in real time based on user behavior.
Build custom memory engines—user embeddings, experience graphs, emotional scoring—and APIs that adapt dynamically.
Collaborate with product and AI teams to refine AI behavior based on emotion logs, memory history, and user feedback.
Own the personalization logic across the system.
You’re Probably Right for This If:
You’ve shipped entire production backends at early-stage startups—moving fast while maintaining code quality.
You’ve integrated or scaled LLM-based products, ideally with emotion, memory, or personalization layers.
You care about system design, response times, clean abstractions, and reliable infrastructure.
You’ve led zero-to-one builds and thrive in environments where you can own both the product vision and technical foundation.
You’re proactive, adaptable, and thrive in lean teams—comfortable building things right rather than managing bloat.
Tech Stack
Languages: C# / .NET / ASP.NET, Python (ML Intelligence Microservices)
Datastores: PostgreSQL, Redis/ValKey (cache + pub/sub), Neo4j/Neptune (Graph RAG), S3 Datalake
Cloud: AWS (RDS Aurora, ECS, Lambda, S3, Route 53, CloudFront, IAM, SQS, SNS, SES, etc.)