Software Engineer & Agentic AI Specialist

Building Intelligent Multi-Agent Systems

Architecting production-grade multi-agent architectures, ReAct orchestration engines, and hybrid LLM pipelines that autonomously reason, plan, and execute complex workflows. Specialized in LangGraph, FastAPI, LiteLLM, Pydantic/Instructor, and VectorDB RAG semantic caching.

Let's Connect Explore Projects 📄 Download CV

🤖

Multi-Agent Systems LangGraph & ReAct

⚡

Semantic Caching & RAG VectorDB Optimization

Technical Mastery

Core Competencies

🏗️

Backend Architecture

Designing distributed microservice patterns, high-throughput asynchronous services, and multi-threaded background processing pipelines capable of processing enterprise industrial workloads.

FastAPI Django Flask Microservices Asyncio / Threading RabbitMQ

🤖

AI Systems & LLM Engineering

Building multi-agent reasoning workflows, ReAct orchestration engines, vector-based RAG semantic caching, and deterministic schema-constrained outputs using modern LLM tooling.

LangGraph LiteLLM Pydantic & Instructor VectorDB / RAG ReAct Engines

🗄️

Database & Data Engineering

Developing complex MongoDB aggregation pipelines, high-throughput time-series data logging, industrial IoT/SCADA protocol connectors (OPC UA, MQTT), and automated validation pipelines.

MongoDB Aggregations Time-Series DBs OPC UA / MQTT SQL Nginx & GCP

Career Journey

Work Experience

Software Engineer

🏢 Itanta Analytics

February 2024 – Present

Backend Development Django Flask FastAPI LangGraph Gen AI Google Cloud APIs

Engineered a multi-agent "Prompt-to-Insights" (PTI) analytics platform utilizing a hybrid LLM architecture and a vector-based semantic caching layer with Retrieval-Augmented Generation (RAG) to optimize query consistency, response latency, and code generation reliability.
Designed and deployed a custom ReAct orchestration engine in Django/FastAPI that translates natural language industrial queries into parameterized MongoDB queries, handling complex multi-level aggregations.
Engineered and optimized data logging connectors (transferring MQTT and PLC data to time-series databases) that improved data transfer reliability and significantly reduced latency for real-time manufacturing workloads.

Featured Work

Highlighted AI Projects

🏆 Showcased at Global ProveIt Conference (USA) Jan 2026

Prompt-to-Insights (PTI) Analytics Engine

Gen AI Developer • Multi-Agent System

1. Integrate (Schema Mapping) ➔ 2. Analyze (QA Multi-Agent) ➔ 3. Visualize (Code Gen & Cache)

Designed a multi-module enterprise platform structured into Integrate (data source configuration & schema mapping), Analyze (natural language querying), and Visualize (dashboard management), streamlining end-to-end analytical report workflows.
Engineered an advanced Analyze QA engine driven by a multi-agent LLM workflow: utilizing fast models for intent parsing and parameterized query (PQ) generation, followed by a VectorDB semantic cache lookup to ensure logic consistency against historical queries.
Implemented a robust code generation and validation pipeline where developer agent dynamically writes and refines component-generation code, routing through automated verification stages before execution in a secure sandbox.
Integrated user feedback loops (Thumbs Up/Down sentiment capture) directly into the VectorDB cache storage layer to continuously penalize or promote cached code performance.

LangGraph LiteLLM Django FastAPI MongoDB VectorDB / RAG RabbitMQ Python

⚙️ Full Stack AI Engine Mar 2026

Industrial IoT Tag Harmonization Engine

Full Stack Developer • AI Pipeline Architecture

Raw SCADA Tags ➔ FastAPI & LiteLLM ➔ Pydantic Schema Guards ➔ ISA-95 Hierarchy

Architected a multi-model AI pipeline using FastAPI and LiteLLM to automatically map messy industrial IoT/SCADA tags to the ISA-95 process hierarchy (Enterprise, Site, Area, Work Center, and Process Units).
Designed a high-throughput parallelized processing engine capable of handling bulk file uploads via threaded concurrency, executing multi-stage model generation and cross-validation loops to ensure optimal mapping precision.
Implemented strict schema enforcement and automatic retries using Pydantic and Instructor, paired with a deterministic safety-net mechanism that guarantees zero silent drops for unmapped inputs.
Developed an administrative SME refinement feedback loop that captures expert corrections and dynamically injects structured guidance into future pipeline iterations for continuous rule consistency.

Python FastAPI LiteLLM Pydantic & Instructor MongoDB Nginx GCP Docker

Technical Stack

Skills & Technologies

🤖 AI & LLM Engineering

LangGraph Multi-Agent Systems VectorDB / RAG Pydantic & Instructor LiteLLM Generative AI ReAct Prompting Prompt Engineering

💻 Backend & APIs

Python FastAPI Django Flask RESTful APIs Microservices Asyncio / Threading RabbitMQ

🗄️ Databases & Data

MongoDB (Aggregations) Vector Databases SQL Time-Series DBs OPC UA Connectors MQTT Protocol

☁️ DevOps & Tools

Git & GitHub Linux / System Processes Google Cloud (GCP) Nginx JavaScript (JS) HTML5 / CSS3

Background & Recognition

Education & Achievements

🎓

Academic Background

B.Tech in Artificial Intelligence and Data Science

Vishwakarma Institute of Technology (VIT Pune)

Graduated: May 2024 CGPA: 8.7

Higher Secondary Certificate (HSC)

Science Stream

Completed: March 2020 Score: 88.15%

Secondary School Certificate (SSC)

General Education

Completed: March 2018 Score: 93.60%

🏆

Honors & Speaking

Showcased at Global ProveIt Conference (USA)

February 2025 • International Presentation

Developed an advanced natural language-driven analytics engine and industrial data pipeline for real-time dashboards; recognized as a top AI innovation and presented to an international audience of 500+ attendees.

Continuous Innovation in Agentic AI

2024 – Present

Architecting cutting-edge multi-agent systems and semantic cache architectures for mission-critical industrial manufacturing deployments.