# Rohit Anand > Engineering leader building enterprise platforms, Voice AI, and Agentic AI systems. Real-world case studies, architectural principles, and technical writing on AI and software engineering. This is Rohit Anand's personal engineering portfolio and publication. Articles identify reference designs, worked examples, and engineering perspectives. Illustrative figures and intended benefits are distinct from measured production outcomes. Case studies describe role, architecture, decisions, and evidence limits. ## Main pages - [Rohit Anand](https://rohitanand.dev/): Engineering leadership, enterprise software, and production AI systems. - [Writing and case studies](https://rohitanand.dev/writing): Technical articles, reference designs, and engineering case studies. - [Start here](https://rohitanand.dev/start-here): A reading path through agent failures, evaluation, bounded execution, and voice systems. - [Engineering principles](https://rohitanand.dev/principles): How Rohit approaches engineering decisions and leadership. - [About Rohit Anand](https://rohitanand.dev/about): Professional background, engineering responsibilities, and areas of work. - [Talk with Rohit's AI guide](https://rohitanand.dev/talk-with-rohit): An AI guide grounded in Rohit's public work; it is not a live conversation with Rohit. - [Contact](https://rohitanand.dev/contact): Contact Rohit about engineering leadership, advisory work, and collaboration. ## Technical articles - [AI in SaaS vs SaaS in AI: From Features to AI Coworkers](https://rohitanand.dev/writing/ai-in-saas-vs-saas-in-ai): Engineering perspective. AI in SaaS vs SaaS in AI: how we build AI coworkers around customer outcomes at Birdeye, with lessons on delegation, reliable execution, and human control. - [Designing a bounded agent workflow: states, tools, and recovery](https://rohitanand.dev/writing/how-we-built-production-ready-multi-agent-system-at-scale): Reference design. A booking example that separates uncertain model decisions from authorized actions, with retry limits and explicit recovery. - [Where voice-agent latency goes: a turn-by-turn budget](https://rohitanand.dev/writing/how-we-built-production-ready-voice-agent-1m-users): Practical guide. Define timing boundaries, stream speech carefully, and test interruptions without confusing a latency target with measured performance. - [Five agent failure modes to test before release](https://rohitanand.dev/writing/the-5-most-common-ways-ai-agents-fail): Practical guide. Concrete failure cases for validation, incomplete actions, retries, context loss, and ungrounded answers. - [Debugging an agent that said “done” before the task finished](https://rohitanand.dev/writing/monitoring-debugging-ai-agents-production): Worked example. An explicitly synthetic investigation showing how traces, service records, and response checks reveal partial completion. - [Why voice agents interrupt callers: testing turn detection](https://rohitanand.dev/writing/state-of-voice-ai-2024-demos-to-daily-use): Practical guide. A text heuristic, its limits, and a practical test matrix for incomplete utterances and response delay. - [Build an agent evaluation suite: cases, rubrics, and release decisions](https://rohitanand.dev/writing/practical-guide-evaluating-llm-agent-systems): Practical guide. A worked cancellation example with explicit criteria, a validated scorer, human review, and a release decision. - [A practical framework for reducing cost per successful agent task](https://rohitanand.dev/writing/how-we-reduced-ai-infrastructure-costs-60-percent): Practical guide. Account for routing, retries, caching, and escalations while holding the task-success definition constant. - [Reviewing AI-generated changes without losing engineering ownership](https://rohitanand.dev/writing/engineering-leadership-age-of-ai): Engineering perspective. A review process organized around behavioral contracts, failure boundaries, and maintainability. - [From an answer to a verified action: a shipment-change workflow](https://rohitanand.dev/writing/from-rag-to-agents-building-knowledge-that-takes-action): Reference design. A reference workflow for permissions, current policy, confirmation, and uncertain external outcomes. - [The decisions an engineering leader still owns when AI helps](https://rohitanand.dev/writing/where-humans-win-what-ai-cant-do): Engineering perspective. Responsibility, evidence, and product judgment remain explicit even when models contribute useful analysis. - [One bug with a coding agent: the contract, the mistake, and the review](https://rohitanand.dev/writing/how-i-use-ai-to-ship-faster): Worked example. Use a concrete sentence-chunking defect to practice specifying behavior, finding incomplete fixes, and reviewing a patch. - [Choosing a small AI development stack: contracts, costs, and limits](https://rohitanand.dev/writing/my-ai-toolkit-for-solo-builders): Practical guide. A selection guide with runtime validation and explicit provider boundaries, rather than an unverified inventory of personal tools. ## Engineering case studies - [Modernizing Birdeye's Frontend Platform](https://rohitanand.dev/case-studies/birdeye-react-migration): Migrating from PHP/jQuery to React at enterprise scale—platform transformation across engineering, organization, and business delivery. - [Building an Enterprise Voice AI Platform for 24×7 Customer Engagement](https://rohitanand.dev/case-studies/birdeye-voice-ai): A business transformation case for customer operations—24×7 engagement, lower cost-to-serve, and revenue-facing voice experiences enterprises can trust. - [Building an Agentic AI Platform for Enterprise Workflow Automation](https://rohitanand.dev/case-studies/birdeye-agentic-ai): Enterprise automation through reasoning, planning, and tool execution—not chat. Multi-step workflows with memory, verification, and guardrails. - [DORA Engineering Transformation](https://rohitanand.dev/case-studies/birdeye-dora): Organizational change through measurement—how DORA shifted Birdeye engineering from opinion-driven delivery to continuous improvement at 40+ people. - [Birdeye UI/UX Platform](https://rohitanand.dev/case-studies/birdeye-uiux): Product platform strategy through standardization—shared UI patterns so Social, Insights, and Campaigns felt like one Birdeye, not mismatched products. - [Modernizing Snapdeal's Web Platform for Performance and Scale](https://rohitanand.dev/case-studies/snapdeal): Lead UI Engineer work—redesign, React SPA migration, and performance engineering for one of India’s largest e-commerce storefronts, including m.snapdeal.com. - [AI Video Generation Pipeline](https://rohitanand.dev/case-studies/ai-video): An agentic AI pipeline that turns creative briefs into production-ready videos while keeping humans in control. - [LittleCircle](https://rohitanand.dev/case-studies/littlecircle): An AI parenting companion for busy parents who need calm, practical guidance in under a minute. - [MyMindNook](https://rohitanand.dev/case-studies/mymindnook): A AI companion designed around privacy goals that helps people slow down, reflect, and build healthier habits through conversations that remember what matters. ## Optional - [Full public writing in plain text](https://rohitanand.dev/llms-full.txt): Article bodies, code examples, references, and case-study architecture and decisions. - [XML sitemap](https://rohitanand.dev/sitemap.xml): Canonical indexable page URLs. - [RSS feed](https://rohitanand.dev/feed.xml): Published writing and case-study updates.