Overview
Design, build and operate production AI agents and AI-powered features as part of Sphera's AI Center of Excellence, and help other teams build with AI.
What you'll do
- Design, build, and maintain production AI agents and AI-powered features, including new builds and enhancements to existing solutions.
- Build and sustain agentic systems end to end: task decomposition, tool design, context and memory strategy, orchestration, human-in-the-loop checkpoints, and fallback behavior.
- Build the retrieval and grounding layer: document processing, chunking, hybrid search, and citation, so agents answer from Sphera's own data.
- Deploy and operate AI workloads on Azure AI Foundry: model deployments, endpoints, content safety, private networking, and cost and quota management.
- Use Claude Code as a core engineering tool and set the patterns other teams follow.
- Own evaluation: build eval suites and golden datasets, measure accuracy, latency, cost, and tool-call correctness, and gate changes on results.
- Implement AI guardrails, prompt-injection defenses, and PII handling, working with Platform Engineering on identity and network controls.
- Ship through CI/CD with versioned prompts and agent definitions, plus tracing and token/cost observability on every path.
- Support teams adopting shared AI services: review their designs, troubleshoot integrations, and document reusable patterns.
- Establish KPIs for adoption, quality, and cost-per-outcome, and use them to drive improvement and make the team's impact visible.
- Participate fully in the agile process: own sprint tasks, log time worked, and keep details current.
What you'll need
- 4–6 years of software engineering experience, including at least 2 years building with large language models in production.
- Hands-on experience shipping agentic systems: multi-step tool use, structured outputs, state and memory, error recovery, and the judgment to know when an agent isn't the right answer.
- Working command of Claude or a comparable frontier model platform, including APIs, tool use, and agent SDKs.
- Practical daily use of AI-assisted development tooling, Claude Code in particular.
- Retrieval engineering: embeddings, vector and hybrid search, reranking, and evaluating retrieval quality on its own terms.
- Evaluation and observability for non-deterministic systems: offline evals, human review loops, tracing, and cost instrumentation.
- Strong Python, plus one of TypeScript, C#, or Java.
- Azure AI Foundry or Azure OpenAI: model deployment, endpoint configuration, and content safety controls.
- Azure fundamentals: App Service or Function Apps, Key Vault, Storage, Entra ID managed identities, and Application Insights.
- CI/CD with Azure DevOps: YAML pipelines, multi-stage deployments, and secret management.
- REST and async API design, and integration with enterprise systems of record.
- Databases (SQL and NoSQL) and vector stores; solid Git and code review practice.
Details
- Location: India (remote).
Read the full description and apply on the company’s own careers page.