Senior Software Engineer (Typescript / Backend) - AI/ML

Level: Senior software engineer with 5+ years of production experience, suited to an autonomous professional who can own features from concept through deployment.

Required skills: TypeScript/JavaScript, React, TypeScript or Python backend development, API and service architecture, AI/ML integrations, and strong communication and collaboration skills.

Relevant tasks: Building full-stack AI-powered features, integrating and deploying models through inference APIs, creating user interfaces and data visualizations, connecting databases with AI services, and implementing monitoring, versioning, observability, and evaluation in production.

How would you design a TypeScript API that connects ClickHouse Cloud to an external AI inference service?

I would define a versioned, strongly typed API with clear request and response schemas, authentication, timeouts, retries, and error handling. I would isolate provider-specific logic behind an adapter layer so models and vendors can be changed without affecting clients.

What considerations are important when integrating an AI model into a production system?

Important considerations include latency, throughput, cost, security, model versioning, input validation, failure handling, and observability. I would also establish evaluation metrics and monitor quality and drift after deployment.

How would you implement reliable retries for inference API requests?

I would retry only transient failures, such as timeouts or rate limits, using bounded exponential backoff with jitter. Requests should be idempotent where possible, and retry budgets, circuit breakers, and provider-specific limits should prevent cascading failures.

How would you use ClickHouse for an AI/ML feature involving large volumes of data?

I would use ClickHouse for high-throughput analytical queries, feature retrieval, event storage, and evaluation data. I would design appropriate ordering keys, partitioning, materialized views, and query limits to keep interactive workloads predictable.

How would you integrate vector search or a vector database with ClickHouse Cloud?

I would define an embedding generation pipeline, store embeddings and metadata with consistent identifiers, and expose similarity search through a well-designed service API. The design should address embedding model versioning, re-indexing, filtering, freshness, and fallback behavior.

How would you design observability for an AI inference service?

I would collect structured logs, distributed traces, metrics for latency, error rates, throughput, token or compute usage, and provider availability. I would also track model-specific signals such as confidence, evaluation scores, feedback, and input or output quality while protecting sensitive data.

How would you prevent sensitive data from being exposed to an external AI provider?

I would classify and minimize data before transmission, apply redaction or tokenization, enforce tenant isolation, and use encryption in transit and at rest. Access controls, audit logs, retention policies, and provider data-processing guarantees should be verified before production use.

How would you build a responsive React and TypeScript interface for an AI-powered database feature?

I would model asynchronous states explicitly, including loading, partial results, errors, cancellation, and retries. The interface should provide clear feedback, progressive rendering where appropriate, accessible controls, and visualizations that help users understand both the data and the model output.

How would you evaluate whether an AI-powered feature is working well?

I would combine offline datasets and automated evaluations with production metrics such as task success, latency, cost, adoption, and user feedback. I would use versioned prompts, models, and datasets so regressions can be detected and results reproduced.

How would you design a TypeScript backend service for scalable, multi-tenant workloads?

I would separate request handling, authorization, business logic, provider integrations, and persistence into testable modules. Tenant-aware quotas, rate limiting, isolation, caching, backpressure, health checks, and horizontally scalable stateless workers would help maintain reliability.

ClickHouse is a globally distributed technology company building a high-performance, cloud-native analytical database platform. As a Senior Software Engineer, you’ll develop AI/ML-powered features across inference APIs, backend services, integrations, and intuitive user interfaces, taking projects from concept through production. The stack includes TypeScript, JavaScript, React, Python, APIs, vector databases, and cloud platforms such as AWS, Azure, and GCP; this is a fully remote role open to eligible EMEA countries, with compensation of $147K–$238K plus equity. Apply to shape next-generation data and AI experiences while enjoying flexible time off, healthcare contributions, a home-office stipend, and a globally collaborative culture.