Kognitiv AI
is a Sweden-based startup building
Kog
— an
AI
agent for enterprises and
S(SaaS)
teams. We move fast, and ship real systems: agent harnesses, tooling, deployment, evaluations, and product surfaces.
We’re looking for an intern who wants to go deep on modern agent engineering and production-grade delivery.
What you’ll work on
You’ll pick up real ownership across the Kog stack (depending on your strengths):
- Core agent harness:
orchestration, tool calling, memory, routing, guardrails, multi-tenant patterns
- Ecosystem & tooling:
developer experience, test harnesses, prompts/config pipelines, observability
- Kubernetes (K8s) & deployment:
containerization, GitOps, scaling, multi-namespace isolation, secrets, ingress
- Evaluations:
offline evals, regression suites, quality metrics, cost/latency tracking
- Product/UI:
simple but sharp
User Interface (UI)
for agent workflows, admin views, dashboards
- Apps & integrations:
connectors, APIs, small services that make the agent useful in real orgs
We actively use modern coding assistants:
Cursor
,
OpenAI Codex
,
Claude Code
, etc. You’ll learn how to
actually
pair with them effectively (and when not to).
What we’re looking for
- You can code and ship (school projects count, but show us something real)
- Comfortable with at least one:
TypeScript
,
Python
,
C# (.NET)
- Basic understanding of Large Language Models (LLMs)
and API-driven development
- You like debugging systems, not just writing greenfield code
- Strong communication in English (Swedish is a plus, not required)
Bonus points
- Hands-on
Kubernetes (K8s)
or
Docker
- Experience with evaluation frameworks, testing, or data pipelines
- Frontend experience (React or similar)
- Security mindset (multi-tenant boundaries, secrets, permissions)
What you get
- Real ownership on a real agent product (not “intern tasks”)
- Mentorship + fast feedback loops
- Exposure to production constraints: reliability, cost, latency, safety
- A place where you can build a portfolio that matters