Building production-grade AI agents, LLM-powered systems, and cloud microservices that solve real-world enterprise challenges. 5+ years shipping distributed systems at scale.
Agentic AILLM PipelinesRAG SystemsAWS / GCPKubernetesPython · Go · TypeScript
I’m a Software Engineer and AI expert with 5+ years of experience specializing in agentic AI systems, LLM-powered applications, RAG pipelines, and cloud-based microservices across AWS and Azure. Currently serving as Head of Agentic AI at Jeeva AI in San Francisco, where I architect production autonomous worker platforms and orchestrate 15+ specialized AI workers.
My work sits at the intersection of AI and distributed systems - I architect autonomous agents that solve real-world financial and enterprise challenges. From designing real-time voice AI platforms with sub-second latency to shipping production ML pipelines, I care deeply about systems that are both intelligent and reliable.
Passionate about autonomous AI, multi-agent orchestration, and developer tooling. When I'm not architecting systems, I'm building side projects like Wisperly and AutoStack.
Technical Skills
Agent Frameworks
LangGraphLangChainDeepAgentsComputer Use APIMulti-agent Systems
5+ years building distributed systems, cloud microservices, and AI-powered applications across the US and India.
Jeeva AIApril 2026 - Present
Head of Agentic AI
San Francisco, CA
Architect a production autonomous worker platform using LangGraph state machines and DeepAgents, orchestrating 15+ specialized AI workers with hierarchical planner-executor patterns to execute complex multi-step business workflows end-to-end.
Designed an LLM routing layer that dynamically dispatches requests across Anthropic Claude, OpenAI GPT, and AWS Bedrock models based on task complexity, latency budget, and token cost, reducing inference spend by 42% while improving p95 latency by 30%.
Built secure code-execution sandboxes (Docker + gVisor isolation, ephemeral filesystems, network egress allowlists) enabling agents to safely run untrusted code, with hard CPU/memory caps and per-tenant isolation.
Shipped computer-use agents leveraging Claude's computer-use API to automate browser and desktop workflows, reaching 87% task success rate across 200+ evaluated workflows.
Built RAG pipelines over Pinecone (12M+ vectors) with hybrid BM25 + dense retrieval and cross-encoder re-ranking; established agent observability and guardrails (LangSmith tracing, eval suites, prompt-injection defenses), cutting hallucination incidents by 60%.
Jeeva AIJune 2024 - February 2026
Lead Systems Architect
San Francisco, CA
Designed and maintained backend services in Node.js, C#, and Go powering AI-driven product features, scaling APIs to 5M+ daily requests at 99.9% uptime with model fallback and request batching.
Built CI/CD pipelines on AWS CodePipeline with automated model and prompt evaluation gates, cutting build times by 55% and shifting releases from weekly to daily.
Authored Terraform and CloudFormation IaC across AWS and Azure to provision GPU inference clusters, vector databases, and LLM gateways, reducing environment setup from 2 days to under 1 hour.
Jeeva AIOctober 2023 - May 2024
AI Applications Engineer
San Francisco, CA
Shipped early LLM-powered application features (chat summarization, lead enrichment, intent classification) integrating OpenAI and Anthropic APIs with prompt templates, retry logic, and fallback across 2M+ events/day; lowered infra costs by 25%.
Tuned MySQL and MongoDB schemas, indexes, and queries powering RAG context retrieval, cutting p99 query latency by 70% on the highest-traffic endpoints.
ScriptChain Health LLC.June 2022 - August 2022
Software Engineer Intern (ML)
San Francisco, CA · Medication Recommendation & Readmission Model
Designed and developed a Graph Attention Neural Network (GNN) for medication recommendation, achieving 83% model accuracy.
Implemented Federated Learning models with the NV Flare library, ensuring secure, privacy-preserving distributed model training.
Deployed models on AWS SageMaker, optimizing training pipelines for large datasets hosted in Amazon RDS.
Path India Ltd.July 2019 - July 2021
Software Engineer (Data)
Mhow, India
Designed and deployed scalable backend systems and microservices handling 1M+ daily transactions across toll collection and smart mobility platforms, ensuring 99.9% uptime and low-latency processing.
Built and optimized data pipelines processing large-scale infrastructure and geospatial data, enabling real-time traffic analytics and improving operational decision-making efficiency by 30%.
Developed internal dashboards and automated workflows, reducing manual processes by 40% and accelerating reporting and compliance integrations with government systems.
// 03. projects
What I'm Building
Production-grade AI systems and developer tools - from real-time voice agents to autonomous software platforms.
A production-grade real-time Voice AI Agent platform enabling low-latency, bidirectional voice conversations at scale. Features a full-stack streaming pipeline handling audio ingestion, speech recognition, language reasoning, and speech synthesis with sub-second response times.
Includes a custom ASR pipeline with fine-tuned transformer-based speech models (Whisper-like), multilingual support, dynamic batching, and a multi-agent LLM orchestration layer with contextual memory and tool-calling.
An autonomous development platform that leverages agentic AI workflows to streamline the software engineering lifecycle. Built to explore how AI agents can handle complex, multi-step software tasks with minimal human intervention.
Integrates event-driven architectures with intelligent agent orchestration, enabling teams to automate repetitive development workflows and focus on higher-level problem-solving.
AI-powered, context-aware sales intelligence platform for modern GTM teams. Built agentic search that understands intent and enriches prospecting workflows autonomously - a key product shipped at Jeeva AI.
Sales IntelligenceAgentic AIRAGGoNode.js
// 04. press
In the News
Featured in publications covering agentic AI, enterprise sales technology, and autonomous intelligence systems. Recognized as a thought leader building the next frontier of AI-powered applications.
TechBullion
The Future Belongs to Agentic AI: A Candid Conversation with Anubhav Sharma, Head of Agentic AI at Jeeva AI
