Skip to content
Hello, I am

SANKETSONWANE

AI ENGINEER

Building intelligent systems that reason, retrieve & act.

AI Engineer focused on Agentic AI, LLM systems, RAG, multi-agent architectures, and applied AI research.

Explore my work GitHub

Pune, India · 2026

Ask MeAI Assistant

Hi! I'm an AI assistant for Sanket's portfolio. Ask me anything about his work, skills, or experience.

■ Impact

ENGINEERING
WITH MEASURABLE IMPACT.

Processing-time reduction10 hours → 10 seconds · ARAI / ZEISS INSPECT automation
0.00%
Rogue-node detection accuracyCOEP Tech research · <3% CPU · <50 ms delay
0%
Weekly usersClass-Node realtime quiz platform
0+
State-level events hostedPlus 1 international
0+

■ Selected work

SYSTEMS, NOT SKILL LISTS.

Real systems with real constraints — agentic AI, retrieval, multi-agent orchestration, and realtime infrastructure.

01

Agentic Commerce / AI Payments / FastAPI · 2026

Agentic Commerce Gateway

AI-transactable merchant infrastructure (ACG)

Built ACG, an agentic-commerce infrastructure layer that makes merchants transactable by unknown AI buyers with deterministic money authority, cryptographic user authorization, real Razorpay test-mode payments, failure recovery, interoperability certification, and independently verifiable transaction provenance.

PythonFastAPIGroqRazorpay Test ModeEd25519SQLiteReact + Vite
Explore case study GitHub
  1. DISCOVERYMachine-readable manifest
  2. CATALOG
  3. CHECKOUTServer-authoritative pricing
  4. USER APPROVALEd25519 signed intent
  5. RAZORPAY PAYMENTTest mode + reconciliation
  6. ORDERIdempotent finalizer
  7. FULFILLMENTReservation-backed
  8. AUDIT TRAILSHA-256 hash-chained
Conceptual flow — illustrative example, not production data.
02

Agentic AI / RAG / PostgreSQL · May 2026 — Jul 2026

Municipal Asset Intelligence Copilot

Agentic AI · RAG · Enterprise AI

An Agentic AI-powered intelligence copilot designed to perform accurate multi-step reasoning across millions of municipal asset records. Built during the Agentic AI Engineer internship at EPIC Technology Research.

Intent-aware query routingHybrid RAGPostgreSQLpgvectorOn-prem LLM orchestration

MillionsAsset records reasoned over

Explore case study GitHub
  1. USER QUERY
  2. INTENT DETECTION
  3. QUERY ROUTER
  4. RETRIEVAL LAYERPostgreSQL · pgvector · Hybrid Search
  5. LLM ORCHESTRATIONOn-prem
  6. MULTI-STEP REASONING
  7. ANSWER
Conceptual flow — components limited to what the resume explicitly supports.
03

Agentic RAG / LlamaIndex · April 2026

TattvaQuest

Autonomous Knowledge Retrieval Agent

An Agentic RAG system for multi-document question answering — chunking, embedding-based retrieval, hybrid search, query routing, tool calling, contextual reasoning, and retrieval-quality optimization.

LlamaIndexRAGEmbeddingsHybrid SearchTool Calling
Explore case study GitHub
  1. DOCUMENTS
  2. CHUNKING
  3. EMBEDDING
  4. RETRIEVAL
  5. HYBRID SEARCH
  6. QUERY ROUTING
  7. CONTEXTUAL REASONING
  8. ANSWER
Conceptual pipeline — illustrative example, not production data.
04

Multi-Agent AI / LangGraph / FastAPI · March 2026

QA-Forge

AI-native Multi-Agent Auditing Platform

An Agentic AI multi-agent platform supporting agent collaboration, task orchestration, tool execution, API validation, adversarial testing, and automated engineering reports.

LangGraphFastAPIMulti-Agent AI
Explore case study GitHub
  1. ORCHESTRATOR
  2. AGENT COLLABORATIONTask orchestration · Tool execution
  3. API VALIDATIONAdversarial testing
  4. VERIFICATION
  5. ENGINEERING REPORT
Conceptual collaboration flow — agent roles kept generic; only implementation-backed roles should be named.
05

Realtime / React / Supabase · January 2025

Class-Node

Real-time Quiz & Classroom Interaction System

A scalable real-time quiz platform with synchronized multiplayer participation, AI-generated quizzes, live analytics dashboards, and scheduled assessments.

ReactTypeScriptNode.jsSupabase

100+Weekly users

Explore case study GitHub
  1. HOST CREATES QUIZAI-generated
  2. REALTIME SYNCSupabase · Node.js
  3. MULTIPLAYER PARTICIPATIONReact · TypeScript
  4. LIVE ANALYTICSDashboards · Scheduled assessments
Conceptual flow — illustrative example, not production data.

■ Experience

SHIPPED WITH REAL TEAMS.

Industry engineering, automation, and applied research — three internships across AI systems, industrial automation, and network security.

  1. 2026

    Agentic AI Engineer Intern

    EPIC Technology Research Private Limited · Pune

    May 2026 — July 2026

    Agentic AIHybrid RAGPostgreSQLpgvector

    Built an Agentic AI-powered Municipal Asset Intelligence Copilot using intent-aware query routing, hybrid RAG, PostgreSQL, pgvector, and on-prem LLM orchestration, enabling accurate multi-step reasoning across millions of asset records.

  2. 2025–2026

    Python Developer Intern

    ARAI — Automotive Research Association of India · Pune

    December 2025 — February 2026

    PythonZEISS INSPECT APIAutomation

    Built and deployed a production-ready Python automation pipeline using the ZEISS INSPECT API, automating marker association, validation, and quality checks for industrial photogrammetry.

    10 hrs → 10 secs99.94% processing-time reduction

  3. 2025

    Research Intern — Network Security & 5G Systems

    COEP Tech · Pune

    May 2025 — July 2025

    SDN5G MECPythonMininetRyu

    Developed a lightweight SDN-based rogue node detection framework for 5G MEC, validated through Python, Mininet, and Ryu-based simulations, resulting in a paper submission to AEUE (Q2 Journal).

    98%Accuracy · <3% CPU · <50 ms delay

■ Research

WHERE ENGINEERING
MEETS EXPERIMENTATION.

Paper · Submitted to AEUE — Q2 Journal

Lightweight Trust-Based Detection of Rogue Nodes in MEC

A lightweight SDN-based rogue-node detection framework for 5G MEC, developed as a Research Intern at COEP Tech and validated through Python, Mininet, and Ryu-based simulations.

SDN5G MECPythonMininetRyuNetwork Security
Accuracy
98%
CPU Usage
<3%
Delay
<50ms

NVIDIA Nemotron Workshop

Presented an AI-powered cybersecurity project

  • — AI Engineers
  • — NVIDIA Teams
  • — Industry Professionals

Presented at NVIDIA's Nemotron Workshop. No employment affiliation claimed.

■ Technical stack

SKILLS WITH EVIDENCE.

Every skill below is sourced from the resume. Hover a highlighted skill to see the project that proves it — only relationships the source material supports are linked.

AI / LLM

Backend

Cloud / DevOps

Programming

■ Education

B.E. Computer Engineering

SPPU

2024 — 2027SGPA: 9.14

Diploma in Information Technology

AISSMS Polytechnic, Pune

2021 — 202485.13%

■ Leadership

BEYOND THE CODE.

Training & Placement Coordinator — Head

PVPIT

Led placement coordination and student engagement.

HackerRank Campus Crew Lead

HackerRank

Representing and leading campus-wide coding, community, and technical initiatives.

Event Host

3+ State-level · 1 International

Hosted 3+ state-level events and 1 international event.

■ About

ABOUT

I'm an AI Engineer focused on building intelligent systems around LLMs, retrieval, agents, and backend infrastructure.

My experience spans industry engineering, AI system development, automation, and applied research — from Agentic AI systems working across large datasets to lightweight security frameworks for 5G MEC.

■ Contact

LET'S BUILD SOMETHING INTELLIGENT.

Open to AI engineering, research, internships, and ambitious technical projects.

sanketsonwane512@gmail.com GitHub LinkedIn Download Resume