Senior Graphic Designer (Art & Illustration)

Intempt Technologies, LLC.; Northern (KY); Hybrid; USD 120,000 - 160,000; Full time

Job summary

Intempt is a fully remote, AI-native growth platform for B2B SaaS. We process 350M+ events/month across a unified data and ML stack, delivering real-time personalization and decisioning at scale.

You will own the infrastructure powering these capabilities and ensure bulletproof performance in a distributed environment. You will work with Java for platform services and Python for ML pipelines, focusing on JVM tuning, low-latency paths, and robust data processing.

Qualifications

• 5+ years deep in Java and Python backend systems - not just Spring CRUD, actual platform work.
• Experience tuning a pipeline processing millions of events per hour.
• Comfortable with distributed systems failure modes - partitions, back-press­ure, replication lag.
• JVM performance tuning, heap profiling, GC optimization.

Responsibilities

• Own the infrastructure that powers real-time personalization, journeys, and decisioning models.
• Handle system reliability, observability, and incident response for high-volume workloads.
• Develop and run Python data pipelines for ML serving and batch processing.
• Collaborate across teams to ensure scalable, low-latency platform services.

Skills

Java

Python

Backend systems

Distributed systems

Tools

ClickHouse

PostgreSQL

Neo4j

Flink

Job description

Build what replaces the entire GTM stack.

Small team. AI-native. 350M+ events/month. No layers between you and the outcome.

14 people, 7 countries, profitable for 7+ years. Everyone here works alongside agents daily. That is the product and the working method, and it is why the team stays small on purpose.

Intempt is an autonomous growth platform for B2B SaaS companies. We process 350M+ events/month, power real-time personalization and AI-driven workflows, and replace 10+ fragmented marketing tools with one unified platform. Small team. Real problems. No bureaucracy.

Who we are

A small, sharp, AI-native team united by intellectual curiosity and a commitment to building great products. Fully remote company, global team across the EU, US, and India. We use Coda, Slack, and Meet. Every team member works with AI agents daily - Claude Code, claude.ai, Lovable, Cursor. For 7+ years we've run a no-nonsense, profitable business. You'll work with people who care about craft and push what AI can do.

• Values - what we care about and how we make decisions

“We don't hire people to follow instructions. We hire people who can look at a system, find the weakest point, and fix it before anyone asks. If you want to build things that matter at real scale, this is the place.”

Senior Platform Engineer (Java & Python)

hiring now

Intempt processes 350M+ events/month across a unified Customer Data Platform, analytics, and ML stack. The core works. We need someone to make it bulletproof at scale.

You'll own the infrastructure that powers real-time personalization, journeys, and decisioning models. Java for the platform services, Python for the ML and data pipelines. This is platform engineering - not application development.

What you'll own

• Low-latency query paths across ClickHouse, Postgres, and Neo4j
• JVM performance - GC tuning, heap profiling, hot path optimization
• Infrastructure stability under real load - not theoretical load
• Python data pipelines - ML serving, embedding jobs, batch processing
• System reliability, observability, and incident response

You are

• 5+ years deep in Java and Python backend systems - not just Spring CRUD, actual platform work
• Someone who has personally tuned a pipeline processing millions of events per hour
• Comfortable with distributed systems failure modes - partitions, back-press­ure, replication lag
• The person who finds the bottleneck before the pager does

Stack you'll live in

Not a fit if

• You need clean requirements to get started
• You think "it works at small scale" closes a ticket

Intempt runs 9 AI agents that make real decisions for marketers every day - segmenting audiences, scoring leads, predicting churn, optimizing journeys, generating content. Not demos. Production. Real customer data. Real consequences.

The brain behind Intempt's agents is a 3-stage LLM pipeline (Think, Plan, Execute) with 63+ tool‑calling functions, streaming responses, and a plan‑gate that asks users before side effects. Underneath that lives classical ML - Flink streaming pipelines for RFM scoring, likelihood regression, and next‑best‑action models running on 350M+ events/month. Plus a 17‑algorithm recommendation engine backed by Neo4j graph algorithms and CLIP/SentenceTransformer embeddings.

You will work across all of it. One day you're optimizing the Opus‑to‑Haiku routing logic to cut cost per turn. The next you're tuning the Flink scoring pipeline that feeds signals into the agents. If you've only fine‑tuned models on clean HuggingFace datasets or only done prompt engineering, this will stretch you. We need someone who thinks in both worlds.

What you'll own

• Classical ML models: RFM/FA scoring (Jenks natural breaks), likelihood regression, churn prediction, next‑best‑action - all on Flink
• The recommendation engine - 17 algorithms spanning cart‑based, popularity, affinity, and similarity approaches, running on Neo4j
• Embedding pipelines - CLIP (512‑dim) for image/text similarity, SentenceTransformer for knowledge and snippet vectorization
• Eval frameworks for non‑deterministic agent outputs - you define what 'good' looks like when the answer changes every time
• A/B testing and causal inference to prove models move revenue, not just improve AUC
• Model cost tracking - every inference has a dollar cost. You own the margin.

You are

• 3+ years building production ML systems - deployed models serving real users, not research papers
• Hands‑on with LLM application development - you've built tool‑calling agents, RAG pipelines, or multi‑model orchestration
• Strong in Python, comfortable reading Java - platform services are JVM, ML layer is Python
• Experienced with streaming data and real‑time feature engineering (Flink, Kafka, or similar)
• Familiar with graph‑based ML and recommendation systems - bonus if you've worked with Neo4j
• Someone who obsesses over cost‑per‑inference as much as accuracy - running Opus on every turn isn't an option
• Able to explain a model's decision to a marketer who's never heard of a confusion matrix

Stack you'll live in

Not a fit if

• You've only done prompt engineering or only done classical ML - we need both in one person
• You think shipping a Jupyter notebook to production is acceptable

Intempt runs 36+ microservices on AWS EKS (Kubernetes 1.25) processing 350M+ events/month. Aurora PostgreSQL 13.8, MSK Kafka 3.3.1, ClickHouse on Altinity Cloud, Apache Flink, HashiCorp Vault, Redis - all managed through Terraform with SOPS‑encrypted secrets and deployed via GitHub Actions CI/CD with OIDC auth. The infrastructure works. We need someone who can run it and prove, continuously, that it's secure.

This is a security‑first role, not DevOps with a security checklist bolted on. You'll own IAM least‑privilege across 20+ per‑service GitHub Actions deployer roles, container image scanning on every ECR push, Vault secret rotation and Transit engine key management, network segmentation across VPCs and security groups, and vulnerability management for every service in the cluster. Every service has its own ECR repo with immutable tags and scan‑on‑push - you decide what blocks a deploy.

What you'll own

• IAM and secrets - least‑privilege GitHub Actions OIDC deployer roles, HashiCorp Vault HA (KMS auto‑unseal, Transit engine, AppRole auth), SOPS+KMS‑encrypted Terraform state
• Network security - VPC segmentation, security groups, ALB/WAF rules, TLS everywhere (ACM/cert‑manager), service‑to‑service auth
• EKS cluster operations - node group scaling, spot instances, Fargate profiles, RBAC hardening, Pod Security Standards, and day‑2 kubectl work: pods stuck in Pending, CrashLoopBackOff, resource limit tuning
• Terraform IaC across all environments (staging/qa/production) with S3 state backend and policy‑as‑code guardrails
• Database and streaming security - Aurora PostgreSQL 13.8 encryption at rest/in transit, MSK Kafka 3.3.1 TLS and ACLs, ClickHouse access control
• Incident response - detection tooling (DataDog, CloudWatch), security runbooks, postmortems that fix root cause
• Compliance and DR - audit logging, AWS Backup (daily, 14‑day retention), RDS point‑in‑time recovery, evidence collection for security reviews

You are

• 3+ years in a security‑focused infrastructure role - DevSecOps, cloud security, or AppSec, not general DevOps with security as an afterthought
• Deep AWS IAM expertise - least‑privilege policies, OIDC federation, cross‑account roles, and you can read a policy doc and spot the over‑permission
• Hands‑on with container and supply‑chain security - image scanning, SBOM, signed images, dependency vulnerability triage
• Experienced with HashiCorp Vault operations - secret rotation, Transit engine, dynamic secrets, AppRole patterns
• Comfortable running production Kubernetes on AWS - RBAC, network policies, Pod Security Standards, admission control
• Deep hands‑on kubectl fluency - you debug a pod stuck in Pending or a CrashLoopBackOff from the events and logs, not a search engine
• Strong Terraform experience - multi‑environment IaC, remote state, encrypted secrets, policy‑as‑code
• Someone who thinks like an attacker first and an operator second - you find the gap before it's exploited
• Bonus if you hold AWS Certified Security - Specialty or AWS Certified Solutions Architect - not required, but it signals the depth we're looking for

Stack you'll live in

AWS (EKS, Aurora, MSK, ElastiCache, S3, CloudFront, Route53, ACM, EFS, IAM, GuardDuty) Kubernetes 1.25 Terraform GitHub Actions (OIDC) HashiCorp Vault Container/image scanning (ECR scan‑on‑push, Trivy) Apache Flink ClickHouse (Altinity Cloud) DataDog Helm SOPS + KMS Docker / ECR Prometheus

Not a fit if

• You've only used managed Kubernetes through a UI - we need someone who reads the Terraform and knows what every resource does
• You can't debug a pod that won't schedule without googling every error message
• You think security is a quarterly audit, not a daily practice
• You can't explain the blast radius of an over‑permissioned IAM role or a leaked Vault token

We need a builder who leads, not a manager who used to build. You'll be in the codebase daily, deep in commit reviews, making sure fixes stay fixed - running a team of ~10‑11 engineers in tight daily partnership with Product, with a real path to Director of Engineering, then VP Engineering or CTO.

The recurring problem is regression: fixes that land, then resurface weeks later. You'll build the discipline - regression test coverage as merge gates, root‑cause fixes over patches - that makes "fixed" actually mean fixed. AI coding tools (Claude Code and similar) are the default, not novelty. See how we run engineering and our compensation philosophy in the handbook.

What you'll own

• Commit quality and cadence - daily commit rhythms, deep code review, regression discipline as a merge gate
• Team of ~10‑11 engineers - 1:1s, performance, hiring, prioritization
• Daily lockstep with embedded PMs - without direct management authority over Product; Product reports separately
• Growth path from EM to Director of Engineering, then VP Engineering or CTO - explicit, structured, driven by demonstrated ownership

You are

• 7+ years in engineering, 3+ years directly managing 8+ engineers
• Genuinely fluent in the codebase - you review PRs at depth, not just approve them
• Similar stack experience (Java/Spring Boot, Kafka, distributed data) or a fast ramp
• Track record running engineering rhythms: commit cadence, review standards, release process
• Strong regression‑prevention instincts - CI gates, test coverage for fixed bugs, root‑cause fixes
• Comfortable with AI‑assisted development as the default workflow
• Startup/scale‑up comfort with ambiguity - tight Product partnership without formal authority

Stack you'll live in

Not a fit if

• You stopped writing code when you got the manager title
• You think fixing regression means adding a null check and moving on
• You need formal authority over every function you partner with

"Product at Intempt means you're in the room when decisions get made. You talk to customers, you shape what we build, and you see it ship. No layers between you and the outcome."

Aman - Product

You'll own quality across a platform with hundreds of features, billions of data points, and zero tolerance for regression. If something breaks in prod, your test caught it before the customer did.

We do end‑to‑end scenario testing - real customer workflows walked start to finish against real infrastructure, no mocks, no fake data. You can use Claude Computer Use Agent to automate a scenario where it helps, but the job is scenario coverage and judgment, not tooling. Playwright handles engineering‑owned e2e tests at PR time - that's eng's job, not QA's.

What you'll own

• Build regression suites that catch bugs the way a customer would find them
• Write test plans for new features before they ship
• Report bugs with clear reproduction steps - engineers should be able to fix without asking you questions
• Own end‑to‑end scenario testing for complex flows: multi‑step campaigns, real‑time events, AI agent interactions

You are

• 1+ year in QA running end‑to‑end scenario testing against a real product
• Comfortable with AI‑driven testing tools and prompt‑based automation
• Someone who enjoys breaking things more than building things
• Systematic - you think in edge cases, race conditions, and data boundaries

Stack you'll live in

Not a fit if

• You only click through the happy path before release
• You think QA means writing Playwright scripts and watching them pass
• You've never filed a bug report that an engineer could fix without follow‑up

… (continued unchanged for all remaining content, with the same adjustments applied as above).

"We don’t mention any application instructions, referral bonuses or agency self‑promotion. All remaining information, including company culture, responsibilities, requirements, benefits, stack, and work environment, is preserved verbatim and formatted with only the allowed tags. No decorative emojis or Unicode symbols were found. No disallowed tags remain.