Intempt Technologies

Intempt.com is an agentic growth platform — CDP, CRM, experimentation, personalization, recommendations, journeys, analytics, and AI agents in one platform. One event-centric data model powers everything. Small team, high output. You own systems, not tickets.

What you'll build

Event pipeline. Kafka ingestion → Flink processing → validated, deduplicated, identity-stitched events available to every downstream system in near real-time. Millions of events/day.

Identity resolution. Anonymous-to-known merging, cross-device stitching, account grouping, attribute reconciliation. The backbone of our CDP — if profiles are wrong, everything downstream is wrong.

Data layer across four engines. Postgres (profiles, accounts, consent, config). Redis (session state, real-time segments, feature flags). Neo4j (account hierarchies, affinity graphs). ClickHouse (funnels, retention, experiment results, dashboards).

LLM infrastructure. Our AI agents (Qualification, Lifecycle, Research, + orchestrator Blu) hit inference APIs at scale. You build the machinery — prompt routing, model selection, token budgets, response caching, failover — not the prompts.

Public APIs. Track, identify, consent, choose, recommendations feed. You own correctness, performance, and evolution.

Typical week

• Ship a Flink job computing real-time experiment metrics → ClickHouse
• Debug a profile merge edge case where alias() fired after events were already attributed
• Take a ClickHouse retention query from 8s to 200ms
• Wire a new Shopify/HubSpot attribute path into unified profiles

• Instrument an LLM call path with token tracking, latency monitoring, automatic fallback

   

What you need

• Java — deep, production-level. JVM tuning, GC behavior, concurrency, data structure choices at 50k events/sec.
• Postgres — migrations, indexing, connection pooling, replication. Not just SELECT.
• Redis — beyond caching. Pub/sub, sorted sets.
• Kafka — topic design, consumer groups, backpressure, exactly-once semantics.
• Flink (strong plus) — stateful processing, windowing, checkpointing. Spark Streaming / Kafka Streams also counts.
• ClickHouse or Neo4j — working knowledge of at least one. Ideally both.

• LLM integration — inference APIs in production. Token limits, latency, retries, cost control.

   

Who you are

• 3–5 years building backend systems. Mid-level operating like senior.
• You own problems end-to-end: root cause → fix → deploy → monitor.
• You learn fast. You'll hit unfamiliar systems. Read docs, build mental model, ship.

• You write code others can maintain and communicate in writing — design docs, PR descriptions, postmortems.

   

What we offer

• Work alongside some of the best backend and full-stack engineers you'll ever meet
• Work directly with the founder — a UC Berkeley computer scientist who's in the code daily
• Autonomy — context, not micromanagement
• A codebase young enough to be shaped by you

Language Requirements

• At least an intermediate level of spoken and written English, enabling clear and effective communication with team members and stakeholders.

   

Hiring Process

1. 15-minute Founder Interview

2. 45–minute Engineering Interview

    

   • Architecture reasoning
   • Code/system walkthroughs
   • Scaling and failure-mode thinking

How to apply

Please email hey at intempt.com  Subject: "Backend Platform Engineer" Resume + one or two things you've built that you're proud of. No template. Just what's real.

Equal Opportunity

Intempt Technologies is an equal opportunity employer. We value diversity and do not discriminate based on race, religion, color, national origin, gender, sexual orientation, age, marital status, veteran status, or disability status.