Backend Engineer - AI / ML

We are looking for a senior backend engineer to build and operate the core AI/ML-backed systems that power large-scale, consumer-facing products. You will work on production-grade AI runtimes, retrieval systems, and ML-adjacent backend infrastructure, making pragmatic tradeoffs across quality, latency, reliability, and cost. We are hiring applied AI engineers, not engineering-for-ML profiles. This role is not an entry point into AI/ML. You are expected to already have hands-on experience shipping ML-backed backend systems in production. At Proximity, you won't just build APIs; you'll own critical backend systems end-to-end, collaborate closely with Applied ML and Product teams, and help define the foundations that power intelligent experiences at scale.

Responsibilities:

• Own and deliver end-to-end backend systems for AI product runtime, including orchestration, request lifecycle management, state/session handling, and policy enforcement.
• Design and implement retrieval and memory primitives, end-to-end document ingestion, chunking strategies, embeddings generation, indexing, vector/hybrid search, re-ranking, caching, freshness, and deletion semantics.
• Productionize ML workflows and interfaces, including feature and metadata services, online/offline parity, model integration contracts, and evaluation instrumentation.
• Drive performance, reliability, and cost optimization, owning P50/P95 latency, throughput, cache hit rates, token and inference costs, and infrastructure efficiency.
• Build observability by default, including structured logs, metrics, distributed tracing, guardrail signals, failure taxonomies, and reliable fallback paths.
• Collaborate closely with Applied ML teams on model routing, prompt and tool schemas, evaluation datasets, and release safety gates.
• Write clean, testable, and maintainable backend code, contributing to design reviews, code reviews, and operational best practices.
• Take systems from design, build, deploy, and operate, including on-call ownership and incident response.
• Continuously identify bottlenecks and failure modes in AI-backed systems and proactively improve system robustness.

Requirements:

• Bachelor's degree in computer science, engineering, or a related field, or equivalent practical experience.
• 6-10 years of experience building backend systems in production, with 2-3+ years working on ML/AI-backed products such as search, recommendations, ranking, RAG pipelines, or AI assistants.
• Strong practical understanding of ML system fundamentals, including embeddings, vector similarity, reranking, retrieval quality, and evaluation metrics (precision/recall, nDCG, MRR).
• Proven experience implementing or operating RAG pipelines, covering ingestion, chunking, indexing, query understanding, hybrid retrieval, and rerankers.
• Solid distributed systems fundamentals, including API design, idempotency, concurrency, retries, circuit breakers, rate limiting, and multi-tenant reliability.
• Experience with common ML/AI platform components, such as feature stores, metadata systems, streaming or batch pipelines, offline evaluation jobs, and A/B measurement hooks.
• Strong proficiency in backend programming languages and frameworks and API development.

What matters (non-negotiable):

• Has built AI applications end-to-end, not just pipelines (RAG systems, agentic workflows, chatbots, and internal AI tools).
• Deep understanding of retrieval, including chunking strategies, embeddings, indexing and re-ranking, retrieval failures, and tuning.
• Has owned full RAG pipelines: ingestion, chunking, indexing, retrieval, and generation.
• Experience with agentic frameworks (tool calling, memory, and multi-step reasoning).
• Thinks in terms of intelligence delivery.
• Production mindset (evals, guardrails, latency/cost trade-offs).
• Clear understanding of managed AI services (fallback mechanisms, service reliability, and failure handling).
• Clear understanding of MCPs (Model Control/Management Plans).

Desired Skills:

• Experience with agentic runtimes, including tool-calling or function-calling patterns, structured outputs, and production guardrails.
• Hands-on exposure to vector and hybrid retrieval stacks such as FAISS, Milvus, Pinecone, or Elasticsearch.
• Experience running systems on Kubernetes, with strong knowledge of observability stacks like OpenTelemetry, Prometheus, Grafana, and distributed tracing.
• Familiarity with privacy, security, and data governance considerations for user and model data.
• Expected exposure to Python, LangChain / LangGraph, Vector DBs, LLM APIs, and GCP tooling (plus, not mandatory).
• The candidate must explain retrieval internals clearly, have built something real using RAG/agents (internal or external), and have worked on the application side of ML, not only infra or data plumbing.