Nirav PatelNow at Adamson Systems Engineering

Hi, I'm Nirav

AI Engineer

I build GenAI applications end to end using Python, LangGraph, Amazon Bedrock, Claude, Azure OpenAI and take them from a customer's problem to production.

Portrait of Nirav Patel

based in

TOR

Toronto, ONEastern Time

the tools

  • LangGraph
  • LangChain
  • Amazon Bedrock
  • Anthropic Claude
  • Azure OpenAI
  • MCP servers and tool schemas
  • hybrid search (BM25, dense, RRF)
  • LangSmith tracing
  • RAGAS
  • Python
  • FastAPI
  • AWS (ECS Fargate, Bedrock, IAM)
  • Docker
  • Kubernetes
  • Qdrant
  • PostgreSQL

the work

mcpsum README preview

mcpsum

open source

Stop MCP servers from rug-pulling your AI agent.

A lockfile and runtime reference monitor for MCP tools, like go.sum for the tools your agent trusts. Definitions are served from the mcp.lock you reviewed, so text a server swaps in after approval never reaches the model.

  • Rust
  • MCP
  • AI security
  • Reference monitor
MergeCode README preview

MergeCode

open source

This patch passes tests. Would a maintainer merge it?

Judges AI-generated patches the way a reviewer does: runs your checks, catches the tricks that get code to green (deleted asserts, skipped tests, new deps, CI edits) and returns a verdict with evidence.

  • TypeScript
  • Coding agents
  • Code review
  • CLI
VibeForge README preview

VibeForge

open source

Describe the moment. Get a playlist that fits it.

An agentic playlist generator: a mood analyst, a curator and a critic work through a LangGraph state machine to turn plain language into ten well-balanced tracks. Accepted into the 500-AI-Agents-Projects repository.

  • Python
  • LangGraph
  • FastAPI + React
  • Critic loop
Graph RAG README preview

Graph RAG

open source

Answers from your documents and the connections between them.

Qdrant vector search plus Neo4j graph facts, cited and streamed over FastAPI SSE. Runs fully local on Ollama or on Hugging Face, with a golden-set eval harness and Langfuse tracing.

  • Python
  • Neo4j
  • Qdrant
  • Evals

the path

  1. Sep 2026 – Present

    Agentic AI Specialist @ Adamson Systems Engineering

    Full-time · Port Perry, ONnow

  2. Apr 2025 – Sep 2026

    AI Engineer @ DeepIDV

    Full-time · Toronto, ON

    • Owned the Financial Screening product from design to production for identity-verification and lending customers - parsing raw bank statements into verified income, cashflow trends, and recurring obligations scored into an explainable approve/review/decline verdict, with synthetic-document forensics flagging AI-generated paystubs in 412ms at under 2% false positives.
    • Architected and delivered a four-agent production suite end to end - autonomous verification, compliance adjudication, agent-gateway integration, and deepfake detection - on LangGraph and Amazon Bedrock Agents, with multi-step reasoning over checkpointed shared state, graceful fallbacks, and Bedrock Guardrails enforcing safety policies.
    Show 4 moreShow less
    • Built Claire, a conversational AI copilot in the customer dashboard - querying live session records, risk scores, and monitoring alerts in plain English with tenant-isolated data access, handling decline explanations, anomaly analysis, and compliance narratives, and surfacing distribution shifts and velocity spikes proactively.
    • Delivered grounded retrieval for a customer-facing support agent at scale - RAG on Amazon Bedrock Knowledge Bases with hybrid search (BM25, dense vectors, RRF, Cohere re-ranking), every answer returning its evidence chain and confidence interval - and DeepAI, a natural-language analytics layer turning business questions into structured queries and regulatory-ready exports.
    • Built an MCP server and agent gateway integrating the platform with customer and internal systems as a provider-agnostic tool surface, with multi-provider routing across Claude, Bedrock, and Azure OpenAI making model choice a configuration decision rather than a code change.
    • Deployed and operated production services on AWS - asynchronous FastAPI microservices at 350ms p50 on ECS Fargate with CI/CD - and established the evaluation and observability practice: LangSmith tracing, automated eval pipelines, retrieval regression suites in CI, and drift detection on Prometheus and Grafana; cut mean time to resolution 40% and delivery timelines 20-25%.
  3. Jul 2021 – Oct 2022

    Data Analyst @ Digital Pilot

    Ahmedabad, India

    • Launched Power BI dashboards for real-time marketing, sales, and user-activity insight, driving self-serve analytics adoption across 20+ stakeholders and shifting decision cycles from weekly to daily.
    • Streamlined data collection and reporting in SQL and Python, automating data pipelines to cut manual reporting effort 60-70% and support ad-hoc analysis of campaign and feature impact.
  4. Oct 2020 – Jun 2021

    Software Engineer @ Thingslista Automation LLP

    Ahmedabad, India

    • Built automation systems and internal REST APIs using PLCs, Python, Django, and PostgreSQL to connect factory-floor hardware with internal tools, trimming manual operational work 75%.
    • Developed an end-user training program and supporting documentation for newly built tools, shortening onboarding time 25% and reducing support requests from operations teams.
  5. May 2019 – Oct 2020

    Software Engineer @ byPeople Technologies

    Ahmedabad, India

    • Promoted from intern to Software Engineer; built and maintained live client applications full-stack using Python, Django, React, and Angular in an Agile delivery environment, building interactive UI components against the APIs behind them.
    • Collaborated with senior developers on feature delivery, customization, and testing across concurrent client projects, improving delivery efficiency.

the schooling

  • 2024Data Analysis for Business Decision MakingDurham College
  • 2023Artificial Intelligence Analysis, Design & ImplementationDurham College
  • 2016 – 2020Bachelor of Engineering – Information TechnologyGTU

the signal

Every good AI system starts with a messy problem.

currently into

  • Multi-agent systems
  • RAG & retrieval
  • Evals & observability
  • Guardrails & AI security

Agentic AI Specialist · Adamson Systems Engineering