Artificial Intelligence Engineer

(RAG & Agent Systems) — MERN + Python Experience Required: 2-3 years Location: Work from office: Ahmedabad Salary range – Upto Rs 100k per month. Positions: 01 About us: Saakh is an AI First B2B Fintech startup enabling SMEs with business intelligence. 1) Take data driven lending decisions 2) Automate payment collections 3) Prevent late payments 4) Delegate recoveries of stuck payments 5) Build Digital Reputation Identity of businesses About the Role We're hiring an AI Engineer who can design, build, and ship production‐grade Retrieval‐Augmented Generation (RAG) systems and AI Agents end‐to‐end. You will work across the stack—Python for data/ML services and MERN (MongoDB, Express.js, React, Node.js) for the product surface—to deliver secure, observable, and scalable intelligent features. A key part of the role is building and operating MCP servers (Model Context Protocol) and tooling connectors that expose internal data and actions safely to agents. You're hands‐on with vector databases, LLM orchestration, evaluation, and monitoring. You write clean APIs, ship thoughtful UIs, and automate deployment with CI/CD. What You'll Do - Design & implement RAG pipelines: ingestion/ETL, chunking & metadata, embeddings, hybrid search (keyword + vector), reranking, context caching, freshness & re‐indexing. - Build multi‐step AI agents: tool‐use, function calling, planning/state machines (e.g., LangGraph/semantic‐kernel/AutoGen equivalents), error recovery, memory design. - Develop MCP servers & connectors: define tools/schemas, auth, rate‐limits, multi‐tenant isolation, and logging to safely expose internal systems to agents. - Own MERN product surfaces: React frontends (chat/task UIs), Node/Express APIs, WebSockets/Streaming for token‐level updates, and MongoDB persistence. - Engineer Python microservices: retrieval, orchestration, evaluation, and batch jobs; package as containers; expose fast, typed APIs. - Vector DB operations: design indexes, choose distance metrics/ANN algorithms (e.g., HNSW/IVF), tune recall/latency; manage Pinecone/Weaviate/Qdrant/Milvus/pgvector. - LLMOps, evaluation & observability: establish offline/online evals (RAGAS/TruLens/DeepEval), guardrails, tracing (OpenTelemetry/Langfuse), metrics, and A/B tests. - Security & governance: prompt‐injection defenses, PII redaction, data scoping/RBAC, audit logs, rate limiting, content filters, policy‐as‐code. - Performance & cost: caching, batching, streaming, model selection, autoscaling, token/latency budgets, and cost attribution. - Work cross‐functionally with Product, Data, and Infra to prioritize use cases and ship reliable, testable features on a predictable cadence. Required Qualifications - 2–3 years software engineering (or equivalent depth), with production experience in both: - Python (APIs, data/ML services, packaging, testing with Pytest) - MERN: MongoDB, Express.js, React, Node.js (TypeScript preferred) - RAG & Agents shipped to production: retrieval pipelines, embeddings, hybrid search, reranking, function/tool calling, and multi‐step workflows. - Vector databases: one or more of Pinecone, Weaviate, Qdrant, Milvus, pgvector—schema design, index tuning, and ops. - MCP servers: built/maintained Model Context Protocol servers or equivalent agent‐tool bridges; experience defining tools, auth, isolation, and telemetry. - Cloud & DevOps: Docker, Kubernetes or serverless (Cloud Run/Lambda), CI/CD (GitHub Actions/GitLab), infrastructure as code (Terraform/Pulumi). - Testing & quality: unit/integration tests, load testing, contract tests for tool/agent interfaces, data quality checks for corpora. - Security mindset: data governance, secrets management, least privilege, dependency hygiene. Tech Stack You'll Touch - Backend: Python (FastAPI), Node.js/TypeScript (Express/Nest) - Frontend: React (Vite/Next.js), WebSockets/Server‐Sent Events - Data/RAG: MongoDB, Postgres, S3/GCS; vector DBs (Pinecone/Weaviate/Qdrant/Milvus/pgvector) - Agents/Orchestration: LangChain/LangGraph, semantic‐kernel, custom state machines; MCP servers & tools - Infra: Docker, Kubernetes/Cloud Run, Terraform, GitHub Actions; Redis for cache/queues - Observability & Eval: OpenTelemetry, Langfuse, RAGAS/TruLens/DeepEval, Prometheus/Grafana Are you one of us? Mail your resume at ag@saakh.in Visit us: Website – Https://www.saakh.in