Skip to content
Hatim Patwa
Portrait of Hatim Patwa

Hatim Patwa, SDE-3 · Forward Deployed Engineer

I turn ambiguous client problems into production systems.

I drop into unfamiliar production systems, work out what is actually broken or missing, and ship the integration, data pipeline, or AI infrastructure that fixes it.

Indore, India

Get in touch

02 / About

Who I am

I started in 2021 as a web developer building PHP and MySQL applications, then moved into full-stack work with Angular, Node.js, and SQL at CDN Software Solutions, where I worked directly with clients on requirements and delivery, and built blockchain-related applications with Web3.js.

Since January 2025 I've been an SDE-3 and Forward Deployed Engineer at Systango, working across client production systems in procurement, legal technology, and mining operations. The work is rarely greenfield. I inherit multi-repo systems, figure out how the data really flows, and then build what's needed: ERP integrations, SSO, data reconciliation, database migrations, and MCP servers that give LLM agents controlled, read-only access to production data.

What I care about is systems that are understandable and safe to operate. I like root-cause analysis more than quick patches, and I prefer deterministic, enforced guardrails (schemas, state machines, hooks, audit logs) over hoping a prompt behaves. Outside client work I build my own tools and RAG systems, and I use Claude Code heavily in my daily workflow.

How I work

  • Start from the real problem.

    I map the domain and the data flow before writing code.

  • Own it end to end.

    Design, implementation, release coordination, production debugging.

  • Make it observable.

    Reconciliation tooling, scheduled jobs, and data-quality checks so problems surface early.

  • Constrain AI by design.

    Agents get least-privilege, read-only access with enforced boundaries.

Experience

  1. Jan 2025 – present

    SDE-3 / Forward Deployed Engineer

    Systango

  2. Jul 2022 – Jan 2025

    Software Developer

    CDN Software Solutions Pvt. Ltd.

    • Angular
    • Node.js
    • SQL
    • Web3.js
    • Alchemy SDK

  3. 2021 – 2022

    Web Developer

    Saify Technologies

    • PHP
    • MySQL
    • JavaScript

Credentials

  • Claude Certified Architect: Foundations (Anthropic)

03 / Expertise

What I do

  • Integrations and ERP

    • Microsoft Dynamics 365 Business Central integrations: invoice generation, GRN-to-invoice matching, data sync

  • Backend and multi-tenant SaaS

    • TypeScript, Node.js, NestJS
    • Nx monorepos, microservices, REST APIs
    • Django and FastAPI

  • Data engineering and migration

    • ETL/ELT pipelines and data reconciliation
    • Azure SQL Server to PostgreSQL migration on AWS RDS
    • pg_cron scheduled jobs

  • Identity and access

    • Azure AD B2C SSO
    • Dynamic customer mapping and user provisioning
    • Supplier onboarding

  • AI agent infrastructure

    • Rust-based MCP servers
    • Python orchestration and an internal AI analyst interface
    • RAG systems and the Claude API

  • Cloud and delivery

    • AWS (RDS), Azure, GCP
    • Docker, GitHub Actions CI/CD
    • Playwright

04 / Work

Selected work

Client engagements, anonymized. Written up the way I'd write an engineering note.
  1. 01

    Legal-tech data platform · US law firm client

    median Zoho → PostgreSQL replication lag, diagnosed
    19 days
    of legal matters unfindable by matter number, traced to a defect
    13.8%
    Context
    Multi-repo system: Dagster data pipelines, a Python monorepo web app, and TypeScript ETL/ELT ingesting from Zoho CRM, Zoho Desk, Qntrl, and QuickBooks into MSSQL, PostgreSQL, MongoDB, and Snowflake.
    Problem
    Data was arriving late and incorrectly.
    What I did
    • Diagnosed a 19-day median Zoho to PostgreSQL replication lag.
    • Diagnosed a defect that made 13.8% of legal matters unfindable by matter number.
    • Built reconciliation tooling that improved data quality and observability.
    Stack
    • PostgreSQL
    • TypeScript
    • Python
    • Zoho APIs
    • Snowflake
  2. 02

    Reporting database migration · legal-tech platform

    Context
    Reporting databases running on Azure SQL Server, moving to PostgreSQL on AWS RDS.
    What I did
    • Built migration tooling to move reporting databases from Azure SQL Server to PostgreSQL on AWS RDS.
    • Set up pg_cron scheduled jobs with timezone-correct scheduling (IST to UTC).
    • Set up tunneled access to RDS.
    Stack
    • pgloader
    • Python
    • PostgreSQL
    • AWS RDS
    • Docker
  3. 03

    LLM agents with safe access to production data

    Context
    LLM agents needed to work with production reporting data.
    What I did
    • Built Rust-based Model Context Protocol (MCP) servers giving LLM agents secure, read-only access to production reporting data.
    • Built a Python orchestration layer and an internal AI analyst interface.
    • Design principle: least privilege and enforced boundaries, not prompt-level trust.
    Stack
    • Rust
    • Model Context Protocol
    • Python
  4. 04

    Procurement SaaS · ERP integration and SSO

    Context
    An Nx/NestJS microservices monorepo for a procurement platform.
    What I did
    • Built Microsoft Dynamics 365 Business Central integrations for invoice generation, GRN-to-invoice matching, and data synchronization.
    • Implemented Azure AD B2C SSO, dynamic customer mapping, user provisioning, and supplier onboarding.
    Stack
    • NestJS
    • Nx
    • Microsoft Dynamics 365 Business Central
    • Azure AD B2C
  5. 05

    Mining operations SaaS · equipment management

    Context
    A multi-tenant, microservices-based equipment-management platform.
    What I did
    • Delivered validation, internationalization (i18n), equipment forms, billing, and scheduled-job features.
    • Supported runtime migration and release/branch integration.

05 / Projects

Things I build

  • FabricSearch

    RAG search over fashion CAD sheets. An LLM parses the query into hard pre-filters and vector search runs on the filtered subset, with named text and image vectors and confidence-gated OCR routing (PaddleOCR for printed labels, a vision model for graphic text).

    • FastAPI
    • Next.js
    • Tailwind
    • Weaviate v4
    • PaddleOCR

  • Jarvis

    A personal AI assistant daemon on Ubuntu, built on the Claude Agent SDK: a Unix-socket CLI, declarative YAML recipes, and a strict security model with no shell interpolation, PreToolUse hooks, and an append-only audit log.

    • Python
    • Claude Agent SDK

  • Document processing sidecar

    A Docling-based ingestion service with OCR routing and FastEmbed embeddings.

    • Docling
    • FastEmbed

  • Claude Code skills and agent tooling

    Custom skills: mssql-to-postgres (with SQL-equivalence verification), ktdoc (documentation for zero-knowledge-transfer onboarding), md-to-docx, and a domain-mapping skill that structures a domain into knowledge files for NotebookLM. Plus an orchestrator design for a 6–8 repo system.

    • Claude Code

06 / Skills

Technical skills

Core
  • TypeScript
  • JavaScript
  • Node.js
  • NestJS
  • SQL
Also strong
  • Python (Django, FastAPI)
  • Angular
  • React / Next.js
  • Rust
Data
  • PostgreSQL
  • MongoDB
  • MySQL
  • Redis
  • Azure SQL Server
  • ETL/ELT
  • Data reconciliation
Cloud and delivery
  • AWS (RDS)
  • Azure (AD B2C)
  • GCP
  • Docker
  • GitHub Actions
  • CI/CD
  • Playwright
  • Nx monorepos
AI
  • Anthropic Claude API
  • Model Context Protocol (MCP)
  • LLM agents
  • RAG
  • Claude Code
Practices
  • System design
  • Multi-tenant SaaS
  • Microservices
  • Root-cause analysis
  • Release coordination

07 / Contact

Get in touch

Hiring for a senior backend, forward-deployed, or AI-enabled engineering role? Send me a message, or reach me directly.

hatimpatwa5253@gmail.com