Software Developer

About the Guideline Guideline is a global provider of ad intelligence and media plan management technology, powering the strategy, planning, and management of advertising buying and selling for the world’s leading enterprises. Our solutions deliver the industry’s most comprehensive and timely insights, enabling publishers, agencies, brands, investors, and consulting firms to optimize media performance and drive superior business outcomes. Job Type: Full‑time Description Guideline’s proprietary spend and pricing data represents approximately $200 billion in annual media investment across 65 countries, providing the most complete and transparent view of the global advertising marketplace available today.

In 2026 we are accelerating our investment in analytics and AI‑powered solutions for the advertising and capital markets industries. The Software Developer, reporting to the Chief AI Systems Officer, will design, build, and operate production‑grade AI agents that automate workflows across Guideline’s ad intelligence, media planning, and analytics products. This role owns the full agent lifecycle — from prompt and tool design, to orchestration with frameworks such as LangGraph and the Model Context Protocol (MCP), to evaluation, observability, and safe deployment at scale.

This role sits at the intersection of managed agents and traditional backend software engineering. You will partner closely with product, data science, and security to ship agents that meet a high bar for accuracy, latency, cost, and reliability in a regulated, customer‑facing environment processing $200B+ in annual media spend data. This role is hybrid and requires 2 days per week in our Toronto office.

Key Responsibilities Design and ship multi‑step AI agents using modern orchestration frameworks (Claude, OpenAI Agents SDK, or equivalent), including prompt design, state management, tool calling, and human‑in‑the‑loop control. Build and maintain MCP servers and tool integrations connecting agents to internal services, data warehouses, and third‑party APIs; define clean schemas, error handling, and least‑privilege authorization scopes. Implement retrieval‑augmented generation (RAG) pipelines — ingestion, chunking, embedding, hybrid retrieval, reranking — grounded in Guideline’s proprietary spend, pricing, and media datasets.

Develop offline and online evaluations (LLM‑as‑judge, deterministic checks, golden sets, regression suites) that measure agent quality, tool‑use correctness, task completion, latency, and cost before each release. Instrument agents with end‑to‑end tracing and observability (e.g., OpenTelemetry, LangSmith, MLflow) and operate them in production: monitor drift, regressions, prompt‑injection attempts, and hallucination rates. Apply security and safety controls — input/output filtering, prompt‑injection defenses, sandboxed tool execution, PII handling, data residency — in collaboration with Security and Compliance.

Optimize for cost and latency through model routing, caching, batching, and choosing the right level of agency — deterministic workflow vs. autonomous agent — for each problem. Write production‑quality Python with strong testing discipline; contribute to backend services, APIs, and CI/CD pipelines that host agent workloads. Partner with product, data science, and design to translate ambiguous business problems into well‑scoped agent specifications, success metrics, and rollout plans.

Stay current on the rapidly evolving agent ecosystem and bring back patterns the team should adopt — or reject — with a clear rationale. Benefits Guideline offers full‑time employees a comprehensive benefits package based on location. Some benefits may include, but are not limited to: Health, dental, life, and disability insurance RRSP with company match Paid time off and parental leave Teledoc Health services Employee recognition and referral bonuses Equal Opportunity Employer Guideline is an equal opportunity employer, committed to our diversity and inclusiveness.

We will consider all qualified applicants without regard to race, color, nationality, gender, gender identity or expression, sexual orientation, religion, disability, or age. We strongly encourage women, people of color, members of the LGBTQIA community, people with disabilities, and veterans to apply. Requirements 3+ years of professional software engineering experience shipping production systems, with at least 1 year focused on LLM‑powered or agentic applications.

Strong Python skills, including async programming, type hints, testing, and clean API design. Comfort with Git‑based development and modern CI/CD. Hands‑on experience with one or more agent frameworks (LangGraph, LangChain, OpenAI Agents SDK, Anthropic SDK, CrewAI, AutoGen, Pydantic AI) and provider APIs from at least one of OpenAI, Anthropic, or Google.

Practical experience with the Model Context Protocol (MCP) or equivalent tool‑protocol patterns; ability to design clean tool interfaces and reason about authorization scopes. Demonstrated experience building RAG systems, including vector stores (e.g., pgvector, Pinecone, Weaviate), embedding selection, hybrid search, and reranking. Working knowledge of agent evaluation: designing evals, building golden sets, running LLM‑as‑judge, and interpreting results to make ship/no‑ship decisions.

Familiarity with prompt engineering tradecraft and an empirical mindset — preferring measurement over intuition for agent behavior. Solid grasp of cloud infrastructure (AWS, GCP, or Azure), containers (Docker), and at least one production runtime — Kubernetes, serverless, or comparable. Understanding of LLM security and safety: prompt injection, data exfiltration, output validation, sandboxing, and least‑privilege tool access.

Strong written and verbal communication; ability to write design docs, present trade‑offs, and collaborate across product, data, and security functions. Preferred Bachelor’s or Master’s in Computer Science, Engineering, or a related quantitative field — or equivalent practical experience. Experience operating multi‑agent or hierarchical agent systems (planner/executor, supervisor patterns).

Background in advertising technology, media analytics, or financial/capital markets data. Experience with fine‑tuning, distillation, or open‑weights model deployment (vLLM, TGI, llama.cpp). Open‑source contributions to agent frameworks, MCP servers, or eval tooling. #J-18808-Ljbffr