Full Stack Engineer

About Elimai

• Elimai.ai builds AI-native products that solve mission-critical problems in healthcare and manufacturing by building intelligence pipelines and automation. We're a small founding team shipping real systems for real clients, where one engineer's work visibly moves the business. If you want to build agentic AI in production (not slideware) and own systems end to end, this is that seat.

The Role

• You'll design and ship full-stack, AI-first systems — owning everything from the API layer and data model to the agent orchestration and the UI that surfaces it. This is a builder role with real autonomy: you'll make architectural calls, wire up LLM/SLM workflows, and take features from whiteboard to production. You'll work directly with the founders and AI engineers, with a short path from idea to deployed code.

What You'll Build

• Production agentic workflows — multi-step, tool-using agents that classify, extract, summarize, and act on real-world data (documents, emails, claims, images).
• AI-infused backend services: event-driven pipelines (queues, async workers, webhooks) that route work across LLMs, SLMs, and deterministic logic.
• MCP servers and tools that expose internal systems and data sources to agents in a clean, reusable way.
• Full-stack features end to end — REST APIs, data models, and the React/Next front-ends that make the intelligence usable.
• Cost- and latency-optimized inference: model routing, fallbacks, prompt caching, and picking the right-sized model for each job.

Core Engineering Requirements

• 2+ years shipping production software as a backend / full-stack engineer - Mandatory
• Strong in Python (FastAPI) / MERN; comfortable across the stack (React/Next.js a strong plus).
• Solid REST API design; experience with async processing, queues, and event-driven architecture (SQS, RabbitMQ, Kafka, or similar).
• Databases: PostgreSQL plus a NoSQL/document store (MongoDB, DynamoDB); able to model and index for real workloads.
• AWS + containers: hands-on with core AWS (EC2, S3, Lambda, RDS/DynamoDB), Docker, and CI/CD. Kubernetes a plus.
• You write clean, tested, maintainable code and can debug a production incident calmly.

AI Engineering

• LLM development: prompt engineering, tool/function calling, structured outputs (strict JSON schemas), and streaming.
• Agent SDKs & orchestration: hands-on with at least one modern framework — Anthropic Agent SDK, OpenAI Agents SDK, LangGraph, or equivalent — and multi-agent/tool-routing patterns.
• MCP (Model Context Protocol): building and/or consuming MCP servers to connect agents to tools and data.
• RAG & vector search: embeddings and a vector DB (Pinecone, Weaviate, pgvector, FAISS) with attention to retrieval quality.
• SLM development: fine-tuning, evaluating, and deploying small/open models (Llama, Qwen, Gemma, Phi, etc.) for cost- or latency-sensitive tasks; on-device or self-hosted inference a plus.
• Reliability: guardrails, hallucination control, validation/retry layers, and building evals — you treat AI systems as testable, not vibes.
• Practical instinct for cost/latency trade-offs: model selection, routing, fallbacks, and prompt caching.

Nice to Have

• Experience with healthcare or manufacturing data, or regulated/compliance-sensitive environments.
• Document AI: OCR, layout parsing, vision-LLM extraction pipelines.
• Observability for AI systems (tracing, token/cost monitoring, prompt/version management).
• Contributions to open-source AI tooling or a portfolio of shipped agentic projects.

What You'll Get

• Compensation: ₹15–18 LPA (based on experience), plus meaningful equity (ESOPs) in an early-stage company where your work drives outcomes.
• Work: Hybrid - Coimbatore (Option 1), Remote (Option 2).