Want to know if this job is worth applying to?
Pune, Maharashtra, India
Hybrid
Systems & Workflow Delivery: Build, deliver, and maintain production-grade LLM, agent, retrieval, and machine-learning workflows, product platforms, and features using structured orchestration patterns.
Agent Architecture & Orchestration: Design, define, and maintain agent graphs, MCPs, states, transitions, routing logic, retries, fallbacks, escalation behavior, and tool integrations using established patterns.
Prompt Engineering & Structured Outputs: Develop and refine comprehensive prompting strategies across system prompts, task prompts, tool instructions, response schemas, guardrails, and multi-agent interactions.
Memory & Context Management: Configure and design short-term state, long-term memory, conversation history, summarization, retrieval, and privacy boundaries to determine how context is retained, retrieved, or excluded.
Model Selection & Configuration: Select and configure appropriate models based on task requirements, capability, latency, cost, context length, operational constraints, and reliability.
AWS & Platform Integration: Use AWS AI services and platform capabilities—such as Amazon Bedrock and AgentCore—to build, deploy, and operate AI systems, integrating agents with tools, APIs, databases, and existing application services.
Retrieval & Grounding Strategies: Design robust retrieval and grounding strategies, including chunking, embeddings, ranking, metadata, filtering, and source attribution.
Evaluations & Testing: Create, run, and scale quantitative/qualitative evaluation suites that measure correctness, completeness, reasoning, safety, tool use, latency, grounding, and regression detection.
Debugging, Diagnostics & Observability: Investigate and analyze unexpected agent behavior, evaluation results, and production traces to diagnose whether failures originate in the model, prompt, state, memory, retrieval, tools, or application code. Establish monitoring and observability for AI behavior.
Project Ownership & Technical Leadership: Own AI projects from technical design through production deployment and ongoing improvement. Define reusable patterns and libraries that allow software engineers to build on the company's AI harness safely and consistently.
Software Development & Standards: Develop, test, debug, and maintain supporting application code and backend services while applying high standards for security, privacy, testing, and responsible AI.
Collaboration & Guidance: Collaborate closely with product, design, data science, and engineering partners to translate ambiguous problems into reliable AI capabilities, providing technical guidance and code reviews for engineers building AI-enabled features.
1–4 years of experience in software engineering or related fields, including 2+ years with AI, data science, or related technologies.
Strong proficiency in Python, JavaScript, TypeScript, or a similar programming language.
Strong working knowledge of Node.js and JavaScript or TypeScript.
Demonstrated experience building and operating LLM or agentic systems in production.
Strong understanding of prompting, tool calling, structured outputs, agent orchestration, state management, context construction, and memory.
Experience designing agent graphs, multi-step workflows, routing strategies, and failure-handling behavior.
Experience selecting and evaluating models for different tasks and understanding quality, latency, cost, and reliability tradeoffs.
Practical experience with Amazon Bedrock, AgentCore, or equivalent cloud AI services.
Experience with RAG systems, vector databases, embeddings, ranking, and grounding techniques.
Experience building quantitative and qualitative evaluations for LLM and agent behavior.
Experience using traces, logs, production data, and evaluation results to diagnose AI system failures.
Strong understanding of MCPs, APIs, distributed services, databases, cloud platforms, containers, CI/CD, and observability.
Strong system design, communication, ownership, and collaboration skills.