Software Engineer

Who you are

• 2-5 years of software engineering experience
• Polyglot fluency: Strong in at least one modern backend language (Python, Java/Kotlin/Scala, C++, Go, Rust) and comfortable picking up others as needed. We don't care which language you started in — we care that you can pattern-match across them
• Full-stack range: Comfortable working across backend, data/runtime, and at least dipping into frontend or infra when the problem calls for it. Specialists who refuse to leave their lane are not a fit
• CS fundamentals: Data structures, algorithms, concurrency, REST/HTTP, and a working mental model of distributed systems
• High agency, high urgency, high ownership: You don't wait for a perfectly groomed ticket. You diagnose, you decide, you ship, and you communicate. You treat production issues as your problem regardless of whose code it was
• The "builder" spirit: A GitHub repo, side project, technical blog, or open-source contribution that shows you love to tinker and learn
• AI-native: You already use AI tools (Claude Code, Cursor, agentic workflows) as a multiplier on your own work, and you have opinions about where LLMs help and where they don't
• Global collaboration window: Ability to overlap with evening PST working hours for syncs, design reviews, and collaboration with our India/Europe-based leadership and engineering teams
• Experience with Ray, Arrow, Polars, DuckDB, or other modern Python-native data stacks
• Hands-on experience building with LLMs - RAG pipelines, agentic systems, tool/function calling, evals, or MCP servers
• Exposure to Kafka, JVM stacks (Java/Kotlin/Scala), Snowflake, Databricks, or Spark - useful context for our existing platform, but not required
• Experience with Docker or Kubernetes
• A background in competitive programming or contributions to open-source projects

What the job involves

• We process 500+ billion rows daily, and we are now building the next generation of our platform around AI-native connectors, distributed Python runtimes (Ray), and LLM-powered data workflows. We operate with the intensity of a seed-stage startup and we aren't looking for "cogs in a machine." We are looking for builders who move fast, own outcomes, and don't wait to be told what to do
• As an early-career engineer at Nexla, you will write core code, debug production systems, and help us ship intelligent, context-aware connectors and agentic workflows that power the GenAI revolution
• Build AI-native connectors (70%): Implement intelligent connectors that combine traditional integration primitives (pagination, schema evolution, rate-limiting, retries) with LLM-driven capabilities (semantic understanding, agentic recovery, natural-language tool design). You own your code from the first line to the final deployment
• Work with LLMs in production: Design prompts, tool schemas, evaluation harnesses, and guardrails for LLM-backed features
• Build on distributed Python runtimes: Dive into Ray, Arrow, and modern data-processing stacks (Polars, DuckDB) as part of our agentic runtime initiative. You'll learn how to tune them for throughput, latency, and cost
• Solve "dirty" data problems: Real-world data is messy. You'll build self-recovery mechanisms, agentic probes, and automated retries that keep massive pipelines running without human intervention
• Work across the stack: Move fluidly between backend services, runtime code, agent orchestration, and the occasional frontend touch-up. We don't believe in narrow swim lanes
• Architectural growth: Work directly with our CTO and senior leads to understand why we make certain tech choices and how to design for multi-tenancy, low latency, and AI-native workflows
• Documentation & quality: "Done" means documented. You'll write SDK docs and RFCs so the rest of the platform and our customers can build on what you ship
• Unmatched scale: Your code will process billions of rows for global brands on day one
• Direct mentorship: Work in a small, elite team with direct access to senior leadership
• AI-first engineering: We aren't just "using" AI; we are building the infrastructure that makes AI possible for the enterprise — connectors, runtimes, and agentic workflows