graph by datavruti
mygraph.id/lucky

Lucky Patel

AI Engineer

Ahmedabad, Gujarat, IndiaBuilding since 2023 (~4 yrs)
LinkedInGitHub
Headline

My work spans enterprise AI at scale, like a chatbot serving 100,000 daily users, to founding a multi-tenant SaaS end-to-end, owning everything from architecture to infrastructure audits.

5tenants onboarded at launch

SME recruitment agencies

1,00,000daily users on shipped chatbot platform
~99%peak query amplification cut via ISR caching
3,000+candidates processed within 1.5 months of launch
Represented Roles

Specialized Functions & Roles.

Engineering capabilities and specialized roles demonstrated across real production work.

Founding AI Engineer

Primary Role

AI/ML Engineer

Full Stack AI Product Engineer

RAG/LLM Systems Engineer

Signals

Live in Production.

Each signal ties back to a specific, evidenced moment.

Built

9 moments

Systems shipped end-to-end, from schema to deploy.

  • •Launched graph, a professional-identity platform for data, AI, and analytics professionals, built end-to-end with Next.js, Prisma, and a Claude-powered resume and graph generation pipeline.
  • •Led Graph end-to-end as the sole technical owner, designing and building the entire platform while owning UI/UX, database architecture, DB operations, security, audits, and overall system delivery.
  • •mybotgenie has 1,00,000 daily users across all chatbots

Solved

3 moments

Gaps closed - false positives, edge cases, dead ends.

  • •Reduced peak database query amplification by ~99% on viral profile traffic via 60s ISR edge caching and connection pool fixes
  • •Conducted a pre-launch infrastructure load audit on Next.js 15 App Router and Vercel.
  • •Cut deployment time 22% with Webpack-optimized CI/CD

Scaled

1 moment

Taken from prototype to real production volume.

  • •Scaled ADTS to 5K+ driving tests a month at 99% uptime

Led

1 moment

Architecture, mentorship and decisions owned solo.

  • •Led cross-team delivery as informal tech lead on enterprise AI projects
Stack

Daily tools.

What Lucky actually reaches for, pulled from the tags attached to captured moments.

Work

Key Projects By Role

Case studies, not bullet points - impact numbers attached to each one.

datavruti

Founding Engineer · 2025 - Present
Dec 2025

estana - AI-Powered ATS & Recruitment Pipeline

End-to-end ATS for SME recruitment agencies with LLM-based resume parsing and semantic candidate matching; 3,000+ candidates processed within 1.5 months of launch.

Processed 3,000+ candidates across active agency pipelines within 1.5 months of launchOnboarded 5 SME recruitment agencies as tenants at launch.
Stack:Next.jsNode.jsPostgreSQL/PrismaDigitalOceanMulti-tenant SaaS
View case study
Aug 2026 - Sep 2026

graph

Next.js 16 + React 19 + Prisma 7 + Postgres (Supabase), LLM-driven capture pipeline. Shipped v1 solo in 21 days — auth, profile/timeline engine, admin tooling; execution end-to-end mine, product direction from a partner. 100 signups in the first 2 days. Building toward profile discoverability inside AI tools (ChatGPT, Gemini, Claude) and a consent-gated recruiter/agency unlock — paid access to verified AI/data-analytics professional profiles.

Cut peak database query amplification by ~99% (from 6,000-7,000+ queries/min uncached to a fraction of that with ISR + dedup)Eliminated Postgres connection pool exhaustion under horizontal lambda scaling by capping Prisma adapter connections and offloading reads to the poolerPrevented a launch-day outage and silent 100-200 credit loss per failed generation by raising the serverless execution ceiling from 15s to 60s before public launch
Stack:Next.jsPrismaClaudeNext.js 15VercelPostgreSQLISRReact
View case study

Silvertouch Technologies

Software Engineer (AI/ML) · 2023 - 2025

IPSA REC App

React Native + geotagging app; 30% user adoption increase, 15% reduction in distribution losses.

View case study
Jan 2023

Entity Verification System

Real-time multi-entity verification; 60% accuracy increase, 50% faster verification.

60% accuracy increase50% faster verification
Stack:VerificationReal-time systems
View case study
Jan 2023

ADTS (Advanced Driving Test System)

Next.js + ML camera-based test analysis. 25% accuracy improvement, 35% faster data processing, 5K+ tests/month at 99% uptime.

25% accuracy improvement35% faster data processing5K+ tests/month at 99% uptime
Stack:Next.jsComputer VisionMLUptime
View case study
Jan 2023

SAP Agentic AI

Intelligent SAP system with autonomous workflow execution, NL-to-SQL via Vanna AI, BGE+Qdrant retrieval. 95% forecast accuracy, 40% efficiency gain, 20% cost reduction.

95% forecast accuracy40% efficiency gain20% cost reduction
Stack:Vanna AINL-to-SQLBGEQdrantAgentic AI
View case study
Jan 2023 - Aug 2026

MyBotGenie

RAG-based smart chatbot using LLaMA 3.2 8B/vLLM, BGE + Qdrant; 500K+ embeddings indexed, 95% retrieval accuracy, <800ms latency, 10K+ queries/day, 50% recruiter efficiency gain.

500K+ embeddings indexed95% retrieval accuracy at <800ms latency10K+ queries/day50% recruiter efficiency gain
Stack:RAGLLaMAvLLMQdrantBGE Embeddings
View case study
Outside Work

Side projects, open source & publications.

Independent builds, open-source contributions, technical writing, and experiments beyond the day job.

Side project

AI Expense Manager

Next.js + FastAPI + LangChain + OpenAI app for transaction categorization, forecasting, and pgvector semantic search.

Learning

Formal record.

  • MERN Stack Development
    Weltec Institute, Vadodara
    2022
  • Self-directed Learning, Full Stack Web Development (MERN)
    2019 - 2022
  • B.Tech in Information Technology
    Parul University, Vadodara
    2015 - 2019
Activity heatmap

A record of momentum.

A combined view of Lucky's activity across the tools used daily.

LessMore

Profile last updated today.

Liked Lucky's graph?

Build yours today.

Start with your resume