Bounteous The Lead Forward Deployed Engineer (FDE) at Bounteous leads technical design and delivery of agentic AI solutions on Claude. They embed with client teams to build production-grade agentic systems and mentor engineers across geographies.
Responsibilities
Technical Leadership & Architecture
Own the end-to-end technical design of agentic solutions on Claude, from discovery and workflow mapping through production rollout and continuous improvement.
Match autonomy to the use case using Anthropic's agent design patterns: start with the simplest composable workflow (prompt chaining, routing, parallelization) and move to orchestrator-subagent or evaluator-optimizer designs only when the task requires it.
Select the right runtime for each workload: the Claude Agent SDK for agents embedded in client applications, Claude Managed Agents for long-running hosted agents, and Claude Code for engineering workflows, deployed through the Claude Platform, Claude Platform on AWS, or Amazon Bedrock based on client cloud and compliance needs.
Design the tool and context layer: build and govern MCP servers that connect Claude to enterprise systems and data, package domain expertise as Agent Skills, and apply context engineering (progressive disclosure, compaction, memory, subagent context isolation) so agents stay accurate across long, multi-step tasks.
Lead model strategy across the Claude family, matching Opus, Sonnet, and Haiku to task complexity, latency, and volume, and using prompt caching, batch processing, and extended thinking and effort controls to keep cost predictable.
Define human-in-the-loop checkpoints, permission scopes, and approval gates matched to each workflow's risk profile.
Delivery & Engineering Excellence
Lead and mentor engineers across geographies, including Context Engineers, Evaluation Engineers, and Orchestration Architects, through an AI development lifecycle.
Build with Claude Code as the primary engineering environment, using subagents, hooks, plugins, project memory conventions, and headless runs in CI to deliver consistently at speed.
Make evaluation the backbone of delivery: define success criteria before building, create eval suites from real client tasks (code-graded, model-graded, and human review), and gate every release on them.
Instrument observability and cost controls from day one, with tracing of agent sessions, tool calls, and token spend per workflow.
Qualification
Strong prompt and context engineeringProficiency in Python and TypeScriptEnterprise data integration experienceA strong grasp of agent governanceExperience leading teams in complexAnthropic certificationClaude deployments on AWSExperience in globalWhat Success Looks Like
Required
8+ years in software engineering and solution architecture, with significant client-facing or forward-deployed delivery.
A track record of shipping LLM-based agents to production (not only prototypes), ideally on Claude.
Hands-on depth with the Claude API (tool use, structured outputs, prompt caching, streaming), the Claude Agent SDK, and MCP server development.
Daily fluency with Claude Code and agentic coding practices.
Strong prompt and context engineering: system prompts, tool descriptions, and skills that hold up against messy enterprise data.
Experience designing evaluation frameworks and using them to drive engineering decisions.
Proficiency in Python and TypeScript, modern backend architecture, and cloud deployment (AWS preferred).
Enterprise data integration experience, including when to use retrieval-augmented generation versus agentic search over live systems.
A strong grasp of agent governance, security, observability, and risk controls.
Experience leading teams in complex, regulated enterprise environments.
Preferred
Anthropic certification, such as Claude Certified Architect (Foundations or Professional).
Prior experience in BFSI or other regulated industries.
Production experience with Claude Managed Agents or multiagent orchestration.
Claude deployments on AWS, including IAM, network isolation, and data residency requirements.
Experience rolling out Claude Enterprise or Claude Cowork with admin-managed skills and plugins.
Experience in global, distributed delivery models.
Contributions to the MCP or Agent Skills ecosystems, or other open-source work.
