We're looking for a Senior Software Engineer to join our team in Toronto, focusing on building and optimizing state-of-the-art LLM-powered agents that can reason, plan and automate workflows for users. You will be leading the design and development of search and retrieval agent systems that enable users to generate compete insights for their business. In this role, you will own projects end-to-end, guiding architecture decisions, experimentation strategy, and production readiness for LLM-powered retrieval and generation workflows. You will shape how we integrate retrieval-augmented generation (RAG), dense retrieval, query understanding, and agentic reasoning loops to deliver fast, accurate, and trusted search experiences at scale. What You'll Do • Build and ship backend systems that power agentic workflows. You design retrieval pipelines, orchestration layers, and multi-step agent architectures that turn millions of competitive data points (news, press releases, webpage changes, Slack posts, emails, reviews, CRM data) into actionable intelligence for our customers. • Own evaluation of agentic systems at scale. You develop and operate evaluation frameworks (automated, offline, and human-in-the-loop) that measure relevance, quality, latency, and end-to-end task success across our agent pipelines. You'll define what "good" looks like and build the infrastructure to measure it continuously. • Design and optimize retrieval and ranking systems. You work across hybrid retrieval, re-ranking, query rewriting, and post-retrieval synthesis to ensure our agents surface the right information at the right time. You understand the tradeoffs between BM25, dense retrieval, and hybrid approaches and know when each matters. • Improve LLM-powered workflows end to end. From prompt design and retrieval strategy to caching and latency optimization, you'll make our agent responses faster, more accurate, and more reliable in production. • Ship with the customer in mind. You connect technical decisions to customer outcomes. You're energized by understanding how customers use the product, and you use that context to prioritize what to build next. You ship iteratively, measure impact, and course-correct quickly. • Collaborate across product, infrastructure, and data teams - align technical direction with product goals, contribute to architecture decisions, and help the team move faster by establishing patterns and best practices for production-grade agentic systems. • Stay on the frontier. Evaluate and integrate advances in LLMs, retrieval architectures, and agentic reasoning. You have strong opinions (loosely held) about where this space is heading and bring that perspective to your work. What You Bring • Experience building and operating backend systems in production, with meaningful experience in at least one of: search/retrieval, data pipelines, distributed systems, or API-heavy service architectures. • Hands-on experience with search, retrieval, or ranking systems. You've built or significantly improved retrieval pipelines and understand information retrieval fundamentals (hybrid retrieval, relevance tuning, query understanding). • Experience building or evaluating agentic / LLM-powered systems. You've worked with retrieval-augmented generation, multi-step agent workflows, or similar architectures and have thought critically about how to evaluate their output quality at scale. • Strong software engineering fundamentals. You write clean, maintainable, well-tested code. You're comfortable with Python and have experience with backend frameworks, APIs, and production infrastructure. You care about reliability, observability, and CI/CD. • Familiarity with vector databases and search infrastructure. You've worked with tools like FAISS, PGVector, Pinecone, Weaviate, Elasticsearch, or OpenSearch and understand their operational tradeoffs. • Experience with cloud infrastructure (AWS, GCP, or Azure) and building systems that handle scale, large data volumes, low-latency requirements, and high availability. • You use AI coding tools to accelerate your own work. You've integrated tools like Copilot, Cursor, Claude Code, or similar into your development workflow and can speak to how they've changed the way you build software. • Customer-oriented mindset. You've shipped features where you understood the end-user problem, not just the technical specification. You're motivated by customer impact, not just technical elegance. • Ability to lead projects and provide technical direction. You can own a problem end to end, make sound architectural decisions, and help others on the team level up. Nice to Have • Experience designing multi-agent systems or complex orchestration workflows. • Background in conversational search or dialogue systems. • Contributions to open-source projects in search, retrieval, or the LLM ecosystem. • Interest in sharing learnings externally (blog posts, talks, open-source contributions). What Success Looks Like We're looking for builders who: • Take ownership and run with ambiguous problems • Jump into new areas and rapidly learn what's needed to deliver solutions • Bring scientific rigor while maintaining a pragmatic delivery focus • See unclear requirements as an opportunity to shape the solution Our Tech Stack • LLM platforms: OpenAI, Anthropic, open-source models • ML frameworks: PyTorch, Transformers, spaCy • Search/Vector DBs: Elasticsearch, Pinecone, PostgreSQL • MLOps tools: Weights & Biases, MLflow, Langfuse • Infrastructure: Docker, Kubernetes, GCP • Development: Python, Git, CI/CD We encourage candidates to apply to the engineering role and level that best align with their experience. To support a fair and consistent review process, candidates are limited to one engineering application every 60 days. Applying to multiple roles will not improve consideration, as our team evaluates candidates holistically across our engineering career framework throughout the interview process. Not ticking every box? That's okay. We take potential into consideration. An equivalent combination of education and experience may be accepted in lieu of the specifics listed above. If you know you have what it takes, even if that's different from what we've described, be sure to explain why in your application.