Principal AI Systems Engineer

<p><span style="font-family: helvetica, arial, sans-serif;"><strong>About Us: </strong>Founded in 1972, Atari is one of the world’s most iconic consumer brands and a pioneer in the video game industry, known for creating classics like Pong, Asteroids, and Centipede. Today, Atari Inc. continues to build on its legacy by developing games, hardware, and experiences that honor the past while driving innovation for the future.&nbsp;</span></p> <p><span style="font-family: helvetica, arial, sans-serif;">Over the past two years, we've been building Atari India, a growing team that plays a critical role in supporting our global operations. We're proud of the team we've assembled so far, and we’re just getting started. As part of a lean, high-impact organization, the team in India works closely with colleagues in North America and Europe on projects that move the company forward. Whether you're helping launch a new game, keeping our infrastructure secure, or supporting day-to-day operations, your work here matters. Join us as we continue to grow Atari India and build the future of a legendary brand.</span></p> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>Position: </strong>Principal AI Systems Engineer&nbsp;</span><br><span style="font-family: helvetica, arial, sans-serif;"><strong>Experience: </strong>6+ Years </span><br><span style="font-family: helvetica, arial, sans-serif;"><strong>Location: </strong>Netaji Subhash Place, Pitampura, New Delhi</span><br><span style="font-family: helvetica, arial, sans-serif;"><strong>Employment Type: </strong>Full-Time (Hybrid)</span><br><span style="font-family: helvetica, arial, sans-serif;"><strong>Reports to: </strong>Senior Director of Technology, India<br><br></span></p> <h3><span style="font-family: helvetica, arial, sans-serif;"><strong>About the Role&nbsp;</strong></span></h3> <p><span style="font-family: helvetica, arial, sans-serif;">Architect, build, and own AI systems that automate expert-intensive technical workflows end-to end — from CLI frameworks, MCP servers, and agent tooling through to production deployment, business outcome tracking, and continuous improvement. You solve real business problems with AI, ensure solutions are fully implemented and adopted, and measure whether they are actually working.</span><br><br></p> <h3><span style="font-family: helvetica, arial, sans-serif;"><strong>Responsibilities</strong></span></h3> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>System Architecture</strong></span></p> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Own end-to-end architecture of AI automation systems: workflow decomposition, component communication, human checkpoints, and failure behaviour</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Design and build internal CLI frameworks, reusable libraries, and agent scaffolding</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Author and maintain agent instruction files (SKILL.md, CLAUDE.md, system prompts) and MCP server definitions</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Configure Claude Code and Codex CLI environments: MCP wiring, tool permissions, slash commands, and engineering standards</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Evaluate and document architectural trade-offs across reliability, latency, cost, and maintainability</span></li> </ul> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>Pipeline Development</strong></span></p> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Build production-grade AI pipelines in Python: orchestration, structured prompting, context assembly, schema validation, and retry strategies</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Integrate AI systems with external tooling — version control, build pipelines, SDKs, compliance databases, internal APIs</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Design context assembly: how domain knowledge, runtime state, retrieved documents, and tool outputs compose into the precise input each pipeline stage needs</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Build and operate multi-agent systems: orchestrator-worker patterns, agent memory, structured handoffs, and conflict resolution</span></li> </ul> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>Prompt &amp; Context Engineering</strong></span></p> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Design, version, and maintain system prompts and agent instructions as first-class engineering artefacts</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Own output schema design and prompt regression testing with a maintained ground-truth eval set</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Engineer context windows with precision — balancing accuracy, token cost, and latency through compression and selective retrieval</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Define retrieval requirements — able to identify what knowledge is needed, under what conditions, and at what granularity</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Build and maintain structured runtime knowledge assets: curated document corpora, rule sets, decision trees, and validation reference libraries</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Work with domain experts to translate specialist knowledge into agent behaviour: decision logic, edge cases, and failure modes</span></li> </ul> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>Evaluation &amp; Reliability</strong></span></p> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Build and own the evaluation framework: test suites, regression benchmarks, LLM-as-judge pipelines, and per-stage quality metrics</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Run structured failure analysis and implement targeted fixes across context assembly, orchestration, and tool integration</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Define automation rate as a first-class metric and report on business effectiveness of deployed systems</span></li> </ul> <p><span style="font-family: helvetica, arial, sans-serif;"><strong>Governance &amp; Technical Leadership</strong></span></p> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Implement full audit trails — inputs, tools called, outputs, and human review triggers</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Enforce versioning of all agent instructions and system prompts as engineered artefacts with controlled rollout</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Set the technical standard for AI development across the organization — architecture patterns, eval practices, and quality gates</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Collaborate with engineering, product, and domain teams; engage leadership on roadmap priorities and technical risk<br><br></span></li> </ul> <h3><span style="font-family: helvetica, arial, sans-serif;"><strong>Requirements</strong></span></h3> <ul> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Proven track record of building production AI automation systems from scratch — end-to-end from architecture through deployment</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Hands-on expertise with Claude Code, Codex CLI, Cursor, or equivalent — including MCP server configuration and agent instruction authoring</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience designing and deploying MCP servers and custom tools: tool schema, authentication, and permission boundaries</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience building internal CLI frameworks, agent scaffolding, and reusable libraries that others build on</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience creating internal tooling and automation that measurably improved engineering team efficiency — reducing manual processes and accelerating workflows</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience working with data scientists and domain experts to implement AI solutions that measurably improved team productivity</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Deep prompt and context engineering: system prompts, few-shot design, chain-of-thought, token budget management, and prompt versioning</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Proficiency with LLM orchestration frameworks — LangChain, LangGraph, LlamaIndex, AutoGen, or equivalent</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience building AI evaluation frameworks: test suites, regression benchmarks, LLM-as-judge, and production quality monitoring</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Production Python engineering: modular, testable, well-logged code with proper error handling</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Cloud platform experience (AWS, Azure, or GCP): deploying and monitoring AI workloads with containerisation</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience integrating AI systems with external APIs — tool definition, permission management, and failure handling</span></li> <li style="font-family: helvetica, arial, sans-serif;"><span style="font-family: helvetica, arial, sans-serif;">Experience defining and tracking AI productivity metrics: automation rate, time-to-completion, and human intervention rate<br><br></span></li> </ul> <h4><strong><span style="font-family: helvetica, arial, sans-serif;">Shift Timings: </span><span style="font-family: helvetica, arial, sans-serif;">9AM TO 6PM IST</span></strong></h4> <p>&nbsp;</p>