Principal-level AI Engineering & Technical Leadership

Principal AI
Engineer

Architecting and shipping production LLM fine-tuning, RAG pipelines, and multi-agent AI systems — 6+ years, 15+ products delivered under my technical direction.

6+Years
15+Products Delivered
16Engineers Led
100+Paying Users (NIVA)
AI Core
RAG
Agents
LLMs
APIs

I am a principal-level AI engineer and technical lead with more than six years of experience architecting and shipping production AI platforms — from LLM and agentic systems through to full-stack Django engineering.

At PySquad, I grew a cross-functional engineering team from 5 to 16 people and delivered 15+ products under my technical direction, owning architecture standards, tech-stack decisions, and engineering quality end-to-end.

My hands-on work centers on LLM fine-tuning, RAG pipelines, and multi-agent orchestration — including fine-tuning Llama 3.2 on a Hinglish dataset with LoRA and Unsloth, and setting up local inference with Ollama. I've also published two open-source Django libraries on PyPI.

Key areas of expertise:
- GenAI & LLMs: LLM fine-tuning, RAG pipelines, agentic AI, multi-agent orchestration, prompt engineering, tool and function calling
- Backend & Architecture: Django 5.x, Django REST Framework, FastAPI, Flask, multi-tenant SaaS (shared-schema), white-label platforms, event-driven design, microservice patterns
- Data & Storage: PostgreSQL, pgvector, PostGIS, MongoDB, Redis, Elasticsearch, MySQL
- DevOps & Cloud: Docker, Kubernetes, CI/CD (GitHub Actions, Terraform), AWS, Linux/Ubuntu server deployment
- Technical Leadership: architecture standards, hiring and technical interviewing, mentoring, team scaling, roadmap execution

Currently leading AI engineering at PySquad Informatics LLP, open to principal-level AI / GenAI opportunities.

6+Years Experience
15+Products Shipped
2Companies
5 → 16Team Growth

What I Do

  • AI & LLM Engineering

    Production-grade RAG pipelines, agentic AI, and multi-agent orchestration, including fine-tuning Llama 3.2 with LoRA and Unsloth, and local inference with Ollama.

  • Backend & Architecture

    Scalable Python backends with Django 5.x, DRF, FastAPI, and Flask. Multi-tenant, white-label SaaS platforms with shared-schema tenancy, event-driven design, and microservice patterns.

  • Technical Leadership

    Architecture standards, tech-stack selection, code review, and hiring. Grew a cross-functional team from 5 to 16 engineers and mentored juniors into higher-ownership roles.

  • Data & Cloud Engineering

    PostgreSQL, pgvector, PostGIS, MongoDB, Redis, and Elasticsearch, deployed via Docker/Kubernetes with CI/CD (GitHub Actions, Terraform) on AWS.

Tech Stack

PythonDjango 5.xDjango REST Framework FastAPIFlaskCelery LLM Fine-tuningLoRAUnsloth Llama 3.2RAG PipelinesAgentic AI Multi-Agent OrchestrationPrompt Engineering Tool / Function CallingOllamaGemini OpenAIPySparkPandas PostgreSQLpgvectorPostGIS MongoDBRedisElasticsearch MySQLDockerKubernetes AWSTerraformGitHub Actions Linux / Ubuntu

AI Systems

Flagship production AI platforms & notable contributions

  • Agentic AI 100+ paying users

    NIVA — Aviation & Logistics Conversational AI

    Problem: Aviation and logistics businesses needed intelligent, industry-aware chatbots without months of custom development.

    Production conversational AI platform with a 102-entry use-case library, a four-tier pricing model, and white-label bot personas. I led the architecture and delivery, and designed the internal ISPIA architecture framework behind it.

    PythonLLMsRAG Pipelines Multi-Agent OrchestrationFastAPIWhite-label Theming
    Live Project
  • Multi-tenant SaaS Team of 8 · Used internationally

    MarinaPy — Marina Management & Slip Booking SaaS

    Problem: Marina operators needed a single cloud platform to run bookings, berths, CRM, billing, and boater self-service instead of stitching together disconnected tools.

    Cloud-based SaaS platform for marina management and slip booking covering the full operational stack — interactive marina maps with real-time occupancy, booking and berth allocation, CRM, quoting and invoicing, online payments, yard operations, and a self-service boater app. As technical architect, product owner, and delivery head, I led a team of 8 and own the architecture and technical direction for a product now used by marinas internationally.

    PythonDjangoPostgreSQL PostGISReal-time MapsPayments
    Live Project
  • Multi-module Platform Wallet · Tracking · Reporting

    ZAST — Logistics Management Platform

    Problem: Logistics servers needed a single managed web application spanning wallet, tracking, booking, and reporting instead of separate disconnected tools.

    Django-based, fully managed web application for logistic servers across various modules. ZAST includes a wallet facility, live tracking, booking, and various types of reports, among other services.

    PythonDjangoWallet System TrackingReporting
    Live Project
  • Large-scale Systems Large-scale subscriber database

    Airtel — FTTH (Fiber to the Home) Systems

    Problem: Airtel's FTTH rollout needed reliable Python-side systems to manage a large-scale Indian subscriber database.

    Worked on Airtel's FTTH system, handling Python-side development and delivery ownership across a large-scale Indian subscriber database. Managed server operations alongside the team and was responsible for the Python workstreams and on-time delivery of the components owned.

    PythonLarge-scale DatabasesServer Operations
    Live Project

Notable Contributions

Focused, module-level work on external AI products.

  • AI Module

    Predictores.ai — Predictive Campaign Generation Module

    Built the Predictive Campaign Generation module for a human-supervised decision-intelligence platform that turns business and audience signals into explainable next-best-actions.

    PythonLLMsPredictive Modeling
    Live Project
  • AI Module 3 hrs → ~20 min

    CareNav — AI Document Automation for Insurance Claims

    Built AI-based document handling and process-automation workflows for a hospital insurance-claims platform (ABDM-compliant, FHIR R4), bringing a three-hour manual process down to roughly twenty minutes.

    PythonDocument AIProcess AutomationFHIR R4
    Live Project

Technical Writing

AI engineering deep-dives published on LinkedIn for an audience of 8,500+ followers.

Prompt EngineeringCoDLLM
5 min readMar 27, 2025

Chain-of-Draft (CoD): The Prompting Technique

Prompt engineering has evolved significantly, from simple one-shot prompts to complex multi-step reasoning methods. CoD forces models to produce explicit minimal reasoning drafts before final output, dramatically improving accuracy on complex tasks while reducing token usage.

LinkedIn Newsletter Read Article
MCPAI AgentsAPIs
6 min readMar 2025

MCP (Model Context Protocol) - Going to Replace APIs?

In the fast-evolving landscape of AI and LLMs, the strategies we use to enhance model outputs are changing fast. A deep dive into how MCP enables richer, stateful tool-use for AI agents compared to traditional REST API integrations, and what it means for the future of software.

LinkedIn Newsletter Read Article
Neurosymbolic AIDeep LearningReasoning
7 min readMar 2025

Neurosymbolic AI: Bridging Logic & Deep Learning

Neurosymbolic AI is rapidly gaining attention as the missing link between symbolic reasoning and deep learning. This article explores how combining structured logical rules with neural networks opens new frontiers for reliable, explainable AI systems.

LinkedIn Newsletter Read Article
Reinforcement LearningDRLPython
8 min readMar 12, 2025

Deep Reinforcement Learning with Stable-Baselines3

Deep Reinforcement Learning has revolutionised how AI systems learn complex behaviours, from mastering Atari games to robotic control. A practical walkthrough of building and training DRL agents using Stable-Baselines3 with Python.

LinkedIn Newsletter Read Article
View All Articles on LinkedIn Newsletter

Skills & Technology

GenAI & LLMs

LLM Fine-tuning (Llama 3.2)LoRAUnsloth RAG PipelinesAgentic AIMulti-Agent Orchestration Prompt EngineeringTool & Function Calling OllamaGeminiOpenAI

AI & ML

NLPPredictive IntelligenceTrust & Risk Scoring PySparkPandasModel EvaluationData Engineering

Backend

PythonDjango 5.xDjango REST Framework FastAPIFlaskREST API DesignCelery

Architecture

Multi-tenant SaaS (Shared-schema)White-label Platforms Event-driven DesignMicroservice PatternsSystem & Solution Architecture

Data & Storage

PostgreSQLpgvectorPostGIS MongoDBRedisElasticsearchMySQL

DevOps & Cloud

DockerKubernetesCI/CD (GitHub Actions, Terraform) AWSLinux / Ubuntu Server Deployment

Technical Leadership

Architecture StandardsTech-stack SelectionCode Review MentoringHiring & Technical InterviewingTeam ScalingRoadmap Execution

Resume

Education

  1. Bachelor of Engineering (B.E.)

    Gujarat Technological University (GTU)

Experience

  1. Tech Lead (Python/AI)

    PySquad Informatics LLP 07/2021 to Present India
    • Grew a cross-functional team from 5 to 16 people across backend, frontend, AI, and QA, and delivered more than 15 products under my technical direction.
    • Set the architecture standards and owned tech-stack and framework selection across concurrent AI product engagements.
    • Ran hiring and technical interviews, and mentored junior engineers into stronger, higher-ownership roles.
    • Architected multi-agent AI pipelines and RAG systems, including autonomous SDR agents, trust-scoring engines, buying-signal detection, and knowledge assistants.
    • Designed multi-tenant, white-label SaaS platforms with shared-schema tenancy, module-gated access, and per-tenant theming.
    • Fine-tuned and deployed LLMs (Llama 3.2 on a Hinglish dataset using LoRA and Unsloth) and set up local inference with Ollama.
    • Drove repeat business and higher delivery quality through consistent engineering standards, and contributed to pricing and go-to-market strategy.

  2. Software Engineer

    Treesha Infotech 06/2020 to 06/2021
    • Built and maintained Python and Django backend systems, REST APIs, and third-party and payment gateway integrations.
    • Delivered POS and e-commerce modules, from requirement analysis through to UAT deployment.

Core Skills

  • LLM Fine-tuning (LoRA, Unsloth, Llama 3.2)88%
  • RAG Pipelines & Multi-Agent Orchestration90%
  • Prompt Engineering & Tool Calling88%
  • Python96%
  • Django 5.x / Django REST Framework93%
  • FastAPI / Flask88%
  • Multi-tenant SaaS Architecture90%
  • PostgreSQL / pgvector / PostGIS85%
  • MongoDB / Redis / Elasticsearch82%
  • Docker / Kubernetes82%
  • AWS / CI-CD (Terraform, GitHub Actions)80%
  • PySpark / Pandas80%
  • Team Leadership & Mentoring92%
  • Hiring & Technical Interviewing85%

Portfolio

11 projects across AI/LLM, Data Engineering & Web Applications

  • NIVA [Agentic AI Chatbot Platform]
    View Details
    AI / LLM

    NIVA [Agentic AI Chatbot Platform]

    Aviation & logistics conversational AI platform with 100+ paying users.

    Technical Lead / Architect
  • MarinaPy [Marina Management & Slip Booking SaaS]
    View Details
    Web App

    MarinaPy [Marina Management & Slip Booking SaaS]

    Cloud SaaS for marina management & slip booking, used by marinas internationally.

    Technical Architect / Product Owner / Delivery Head
  • AscentPassport [AI Employment Verification]
    View Details
    AI / LLM

    AscentPassport [AI Employment Verification]

    AI verification platform, 10,000+ verifications processed at ~80% accuracy.

    Technical Lead / AI Architect
  • ZingTMS [White-label TMS]
    View Details
    Web App

    ZingTMS [White-label Transportation Management System]

    White-label, multi-tenant TMS on a shared-schema Django base.

    Architect
  • Airtel [FTTH Subscriber Systems]
    View Details
    Web App

    Airtel [FTTH Subscriber Systems]

    Python-side development for Airtel's Fiber to the Home subscriber systems.

    Python Engineer
  • RynoWallet [Coalition Loyalty Platform]
    View Details
    Web App

    RynoWallet [Coalition Loyalty Platform]

    India's first coalition loyalty network connecting local shops via shared RynoCoins.

    Product Owner & Architect
  • ChargeSavvy [POS System]
    View Details
    Web App

    ChargeSavvy [POS System]

    Backend for a UK/USA POS system active across 700+ locations.

    Python Engineer
  • Zast [Logistics Management Platform]
    View Details
    Web App

    Zast [Logistics Management Platform]

    Full-stack Django logistics platform with wallet, tracking, and reporting.

    Senior Backend Developer
  • TDS Reconciliation [Financial Intelligence]
    View Details
    Data / ML

    TDS Reconciliation [Financial Intelligence]

    PySpark reconciliation engine for multinational purchase/sell orders.

    Senior Backend Developer
  • Predictores.ai [Predictive Campaign Generation Module]
    View Details
    AI / LLM

    Predictores.ai [Predictive Campaign Generation Module]

    Predictive campaign module for a human-supervised decision-intelligence platform.

    AI Engineer (Module Contributor)
  • CareNav [AI Document Automation]
    View Details
    AI / LLM

    CareNav [AI Document Automation]

    AI document automation for a hospital insurance-claims platform (ABDM, FHIR R4).

    AI Engineer (Module Contributor)

Open Source

Libraries, tools & platforms I've built and published for the community

2 PyPI Packages
1 Live Platform
MIT License
Django Primary Framework
django-graphify PyPI v0.1.1
Beta

A zero-LLM knowledge graph for Django projects. It inspects models, views, URLs, signals, serializers, admin classes, middleware, and settings fully offline, and generates interactive HTML, Mermaid, Markdown, JSON, and Cypher graph exports.

djangoknowledge-graphstatic-analysisastvisualization
$ pip install django-graphify
PyPI GitHub
django-compliance-shield PyPI v1.1.0
Stable

A drop-in Django library for global privacy compliance. A single decorator delivers DPDP, GDPR, CCPA, and FCRA compliance with field-level encryption, consent management, DSR tracking, breach notification, and data retention out of the box.

djangoGDPRDPDPCCPAFCRAencryptioncomplianceprivacy
$ pip install django-compliance-shield
PyPI GitHub
CivicIssue Open Platform
Live

Civic issue reporting platform empowering Indian communities to report, upvote, and track resolution of local problems — potholes, street lights, infrastructure — with photo uploads, precise geolocation, and a Telegram bot for instant reporting.

civic-techcommunityindiadjangogeolocationtelegram-bot
Live Site
View all on GitHub

Contact

Open to new opportunities and collaborations

Let's build something together

Open to Principal / Lead AI & GenAI roles, contract work, collaborations, and advisory.

Connect on LinkedIn Send Email

Location

Ahmedabad, India

Submit an Inquiry

Fill in the form below and I'll get back to you within 24 hours.

* Required fields