Senior Software Engineer (Data Search)

• Search at Modash isn’t an internal platform or a support function
• It is one of the core products customers use to discover creators
• The scale is large, the data is messy, and the search intent is often complex
• A customer might be looking for creators in a specific niche, people whose content conveys a certain visual style, or accounts that resemble a group they already know
• Solving that well requires more than adding another filter or calling an LLM API
• We need a senior engineer who can work across the full retrieval system—from data and indexing pipelines to embeddings, ranking, relevance, and low-latency serving—and take ambiguous product problems all the way to production
• You’ll join the specialised Data Search team and work closely with Data Core, Data Insights, product teams, customers, and company leadership
• You’ll have real autonomy, but you won’t work in isolation
• You’ll improve how customers discover creators across 400M+ profiles and billions of media files. That includes retrieval, filtering, ranking, relevance, speed, and the product decisions that connect them
• You’ll build systems that generate and use embeddings from images, video, text, and audio at massive scale—then make those signals useful in a real customer-facing search experience
• You’ll evaluate models and technologies pragmatically, understand tradeoffs around cost, latency, and quality, and move promising approaches from experiment to a reliable production system within weeks rather than quarters
• You’ll help shape the problem, gather requirements, design the architecture, write the code, release it, measure the outcome, and improve it. Senior engineers here own results, not just implementation tasks
• Monday. A customer search is returning technically relevant but unhelpful results. You inspect the retrieval and ranking stages, identify where intent is being lost, and propose a measurable improvement
• Tuesday. Deep-focus time. You build a pipeline to generate multimodal embeddings across a large batch of creator content and test how the new representation affects retrieval quality and cost
• Wednesday. You work with Data Insights on a new in-house datapoint. Together, you agree on its definition, coverage, and data-quality requirements, then expose it in Search to give customers more ways to discover creators
• Thursday. You test a reranking model on a fixed set of real customer queries. You measure how much it improves relevance against the latency and inference cost it adds, then decide what’s ready for production
• Friday. You review production metrics, investigate a relevance regression, and share what you learned with the team. The fix may be in the model, the data, the query logic, or the product itself—you follow the evidence
• We keep meetings purposeful and protect time for deep work. You’ll have a short standup, close collaboration when it helps, and plenty of space to design, build, optimise, and launch

Benefits

• Flexible working hours ⏱
• Unlimited paid vacation time 🌴
• Fully remote working 🏠
• Personal development 🧠
• Ownership 💡

• Worked in a fast-moving product environment. You’re comfortable learning quickly, shipping incrementally, and changing direction when the evidence does
• Communicated clearly across teams. You can explain technical tradeoffs to engineers and non-engineers, give direct feedback, and collaborate without creating unnecessary process
• Designed distributed systems. You can reason clearly about throughput, failure modes, data flow, scalability, and operational tradeoffs
• Shipped products from concept to production. You’ve owned scoping, architecture, implementation, release, measurement, and iteration—not just one layer of the solution
• Worked autonomously on ambiguous problems. You know how to gather requirements, ask useful questions, and turn incomplete context into forward motion
• Built large-scale data or backend products. You have solid experience working with systems where volume, latency, reliability, and cost all matter
• Built LLM-powered or agentic features in production. You understand the practical differences between models and can balance capability against latency and cost
• Bonus points if you’ve worked with multimodal embeddings, vector databases, semantic search, ranking algorithms, model deployment, self-hosted models, or GPU infrastructure
• Curiosity about the creator economy helps too, but we’ll get you up to speed

• Intro chat
• Coding interview
• System design interview
• Team interview
• Culture and alignment conversation with our CEO, Avery Schrader