AI/ML Engineer · Agent Systems · Google Developer Expert (AI & Cloud)
I build production LLM systems: MCP infrastructure, multi-agent orchestration, and RAG pipelines that hold up under enterprise load.
| Metric | What it means |
|---|---|
| 92.5% | Query latency cut on MCP infrastructure, 4s → 300ms across 10+ data sources |
| 30% → 90% | RAG retrieval accuracy on production pipelines |
| 60% | Reduction in LLM compute spend, while raising generation quality 40% |
| 87.5% | Fewer LLM hallucinations after system-level prompt guardrails |
| 87% | Faster solution cycles, 15 weeks → 2 weeks, via a CLI tool wiring backend APIs to Claude and Codex |
| 16 | Production AI systems architected end to end |
| 500+ | Secure API transactions per day, zero downtime |
| 30,000+ | Developers trained across APAC, MENAT and SSA as a Google Developer Expert |
| 3 | Peer-reviewed publications |
Agent infrastructure. MCP servers and clients connecting 10+ enterprise data sources, A2A and AG-UI protocol stacks, and multi-agent orchestration with Google ADK. I care about the plumbing: transport, auth, tool routing, and what happens when a tool call fails at 2am.
RAG that survives production. Advanced text extraction, hybrid retrieval, evaluation harnesses, and cost control. Moving accuracy from 30% to 90% is a retrieval and chunking problem long before it is a model problem.
LLM cost and reliability engineering. Confidence scoring, prompt guardrails, Chain-of-Thought and few-shot tuning for voice AI, and orchestration across Gemini, GPT, Claude and Llama so the right model handles the right call.
| Project | What it does | Reach |
|---|---|---|
| a2ui-multiagent-codelab | Codelab for a multi-agent coding assistant on the 2026 protocol stack: A2A, AG-UI and A2UI wired together | Reference implementation |
| rag-with-gemini | Retrieval-augmented PDF analysis with Gemini, +5% QA accuracy over baseline | 49,770+ views |
| adk_hiring_agent | Multi-agent hiring system on Google ADK: resume parsing, skill assessment, role matching | 23,970+ views |
| task_prioritatizaion_with_adk | Agentic task prioritization built on Google's Agent Development Kit | Open source |
| gemma-on-jax | Running and fine-tuning Gemma on JAX and TPU | Open source |
| gemini-api | FastAPI Gemini chat service, Docker and Cloud Run ready | Deployable template |
Resident Entrepreneur, Technology / AI Engineer · Technology 9 Labs (Dec 2025 to Present) Promoted from AI Engineer at Impulsive Web following the 2025 acquisition. Own AI infrastructure strategy and enterprise delivery: MCP pipelines, voice AI optimization, and full-cycle AI integration for 5+ enterprise clients at 95% client satisfaction.
Machine Learning Associate · Binoloop Inc, Remote (Canada) (Jan 2024 to Jan 2025) Promoted from ML Intern to Associate Software Engineer in the same tenure. Built RAG pipelines and orchestrated Gemini, GPT, Claude and Llama across 8+ client engagements for a 30% accuracy and efficiency uplift.
Google Developer Expert, AI and Cloud (Mar 2024 to Present) Grew program reach from APAC-only to three regions in 12 months, training 30,000+ developers through Google-backed programs, talks and workshops.
- Evaluating Large Language Models for Knowledge-aware Question and Answering · International Journal on Smart Sensing and Intelligent Systems
- Optimizing Energy Management in Smart Grids: A Hybrid Approach to Load Forecasting · PACIS 2025
- A Pragmatic Approach on Adoption of EDA to Make Intelligent Business Decisions · International Journal of Wireless Network Security
M.Tech, Artificial Intelligence and Machine Learning, Symbiosis Institute of Technology (2024 to 2026)
B.Tech, Computer Science Engineering, AI and ML, P. P. Savani University (2020 to 2024)