***CAUTION: PLEASE DO NOT SEND APPLICATION IF YOUR CURRENT RESIDENCE IS NOT IN SOUTHERN FINLAND.***
Join our interesting customer development project team.
As a Software Engineer, Agentic AI Platform, you will design, build, and scale the foundational platform infrastructure enabling enterprise-grade autonomous and semiautonomous AI agents. Leveraging Google Cloud Platform (GCP), Vertex AI, and the Gemini ecosystem, you will develop multi-agent orchestration frameworks, tool-calling pipelines, agent memory systems, and secure runtime environments that power enterprise use cases at scale.
Key Responsibilities
• Platform Architecture & Core Services: Design and implement robust backend services, APIs, and runtime environments on GCP to host, orchestrate, and monitor complex multi-agent workflows.
• Agent Framework & Tooling: Build reusable abstractions, agent coordination patterns, dynamic tool-execution interfaces, and retrieval-augmented generation (RAG) pipelines integrated with Gemini models.
• Scalability & Performance: Optimize latency, token throughput, and compute utilization across distributed agent systems using Google Cloud infrastructure (GKE, Cloud Run, Pub/Sub, Firestore/Spanner).
• Enterprise Governance & Security: Implement strict guardrails, role-based access control (RBAC), evaluation frameworks, audit logging, and human-in-the-loop (HITL) interfaces to ensure safe, compliant execution.
• Developer Experience: Create SDKs, developer tooling, and standardized templates that streamline how application teams prototype, deploy, and evaluate agentic solutions.
Required Qualifications & Tech Stack
• Core Engineering: 4+ years of software engineering experience with strongproficiency in Python, Go, or TypeScript/Node.js.
• Cloud & Infrastructure: Solid experience building on Google Cloud Platform (GCP), including container orchestration (Kubernetes / GKE), serverlessinfrastructure, and event-driven architectures.
• AI & Agentic Systems: Hands-on experience with LLM orchestration, function calling/tool use, agentic frameworks (e.g., Agent Development Kit, LangGraph, AutoGen), and vector search/RAG pipelines.
• Gemini Enterprise Agent Platform (Vertex): Familiarity with Vertex AI Agent Builder, Model Garden, Vertex AI Vector Search, and the Gemini API suite.
• System Design: Proven track record in designing high-throughput, low-latency, resilient distributed systems with strong emphasis on observability (OpenTelemetry, Cloud Monitoring).
As a Software Engineer, Agentic AI Development, you will bridge the gap between core AI platform capabilities and high-impact enterprise use cases. You will design, develop, and deploy production-grade autonomous and semi-autonomous AI agents, transforming complex business workflows into intelligent, multi-step reasoning systems powered by the Gemini ecosystem and Google Cloud Platform (GCP).
Key Responsibilities
• End-to-End Agent Implementation: Build and fine-tune domain-specific agents capable of multi-step planning, tool-calling, reflection, and task execution using Gemini models and Gemini Enteprise Agent Platform.
• Workflow & System Integration: Connect agent architectures with enterprise data sources, APIs, relational/vector databases, and SaaS tools to enable autonomous execution within business environments.
• Grounding & Context Management: Design robust context-retrieval strategies, dynamic prompt templates, long/short-term agent memory stores, and hybrid RAG pipelines to ensure accuracy and low hallucination rates.
• Evaluation & Guardrails: Implement automated evaluation suites, benchmark agent performance against real-world scenarios, and integrate safety filters, input/output validation, and human-in-the-loop (HITL) triggers.
• Production Readiness: Partner with platform and data teams to containerize, deploy, monitor, and optimize agent reliability, latency, and cost in production.
Required Qualifications & Tech Stack
• Core Engineering: 3+ years of software development experience with strong proficiency in Python (async frameworks, Pydantic, FastAPI) or TypeScript.
• Agentic Frameworks & Patterns: Practical experience building reasoning chains, multi-agent collaboration, and tool-augmented workflows using frameworks like Google Agent Development Kit (ADK), LangGraph, AutoGen, or LlamaIndex.
• Cloud & Data Systems: Hands-on experience with GCP (Vertex AI, Cloud Run, Cloud Functions, Pub/Sub, BigQuery) and vector stores (Vertex AI Vector Search, Pinecone, or pgvector).
• LLM Engineering: Deep understanding of prompt engineering, structured outputs, function calling, few-shot evaluation, and token/cost optimization.
• Software Fundamentals: Solid foundation in API design (REST, gRPC), CI/CD pipelines, automated testing, and observability (logging, tracing, and agent step evaluation).
As a Software Engineer, Guardrailing (AI Safety & Alignment), you will design, train, and deploy real-time safety, security, and compliance mechanisms that protect generative AI systems and agentic workflows. Sitting at the intersection of applied machine learning, natural language processing, and production engineering, you will build ML-driven filters, classifiers, and adversarial defense systems using PyTorch to detect and mitigate prompt injections, jailbreaks, data leakage, toxic outputs, and model hallucinations.
Key Responsibilities
• ML-Driven Safety Models: Train, fine-tune, and evaluate specialized NLP and classification models (e.g., lightweight BERT variants, custom transformers) in PyTorch for real-time safety classification, semantic moderation, and PIIredaction.
• Adversarial Defense & Red Teaming: Design automated systems to detect adversarial inputs, jailbreak attempts, prompt injection, and indirect injection vectors targeting LLMs and autonomous agents.
• Low-Latency Runtime Integration: Build high-throughput, low-latency guardrail microservices and middleware pipelines that evaluate inputs and model outputs in milliseconds without degrading user experience.
• Evaluation Frameworks & Metrics: Develop programmatic testing pipelines, benchmark datasets, and red-teaming harnesses to continuously measure guardrail precision, recall, latency, and false-positive rates.
• Grounding & Hallucination Mitigation: Implement semantic consistency checkers, fact-verification classifiers, and retrieval-grounding filters to ensure model outputs remain faithful to source data.
Required Qualifications & Tech Stack
• Machine Learning & NLP: Strong background in applied NLP and Deep Learning, with hands-on experience building, fine-tuning, and evaluating transformer models using PyTorch and Hugging Face.
• Software Engineering: 3+ years of production software engineering experience with strong proficiency in Python (FastAPI, async I/O, optimized inference runtimes like ONNX, TensorRT, or vLLM).
• AI Safety & Guardrails: Practical experience with LLM guardrail architectures, safety frameworks (e.g., NeMo Guardrails, Llama Guard, Guardrails AI), and content moderation systems.
• Adversarial ML & Security: Familiarity with common LLM vulnerabilities (OWASP Top 10 for LLMs, prompt injection, data exfiltration, system prompt leakage).
• Production & Deployment: Experience deploying containerized ML services (Docker, Kubernetes) to cloud environments (GCP/AWS) with robust monitoring and telemetry.
Details
Further info: Jari Viitanen, jari.viitanen()kontraktori.fi
If you are interest, please apply immediately. Also report your ability to start on the next week. Also return the self assessment form alongside your fresh CV.
AI Coach – Agentic AI & Enterprise Adoption
Tieto
AI Solution Developer Trainee
Deloitte
Machine Learning Engineer
Reaktor
Senior Data Engineer, Enterprise Data Platform (2 people)
F-Secure Corporation
Senior Software Developer (Windows/.NET) – Tieto Tech Consulting (m/f/d)
Tieto
Chief Process Engineer
AFRY