01
Applied AI & LLM Integration
Building domain-adapted LLMs, RAG pipelines, and LangChain multi-agent workflows.
[ COLLABORATORS & TECHNOLOGIES ]
TRIOVATIVE
RDC CONCRETE
AWS CLOUD CLUB
DIPEX 2025
SANJIVANI COE
LANGCHAIN
PYTORCH
REACT
FASTAPI
LLAMA 3.2
POSTGRESQL
SAM-2
OPENCV
02
Full-Stack Web Development
Architecting robust applications with React, FastAPI, and PostgreSQL.
03
Computer Vision Solutions
Automated pipelines for detection, 3D reconstruction, and area estimation.
04
Advanced Network Security
Designing RL-enhanced Intrusion Detection Systems for adaptive security.
SELECTED WORK
AI-Accelerated Development Pipeline
BEFORE: High development time and complex enterprise workflows needing automation.
AFTER: Designed multi-agent systems using LangChain/LangGraph, integrated LangSmith for testing, and architected full-stack apps.
-50%
Dev Time (3mo→1.5mo)
100%
Workflow Automation
CV Material Volume Measurement
BEFORE: Manual material volume measurement is time-consuming and error-prone.
AFTER: Fine-tuned SAM-2 & Gemini Vision LLMs; engineered 3D reconstruction pipelines via OpenCV.
< 10%
Error margin
-70%
Manual Time
Tax-LLaMA-Ind: Domain-Adapted LLM
BEFORE: Complex Indian Income Tax laws requiring automated legal advisory.
AFTER: Fine-tuned LLaMA 3.2-8B via QLoRA; built hybrid RAG pipeline with FAISS + pgVector.
High
Citation Accuracy
Precise
Context Retrieval
Food Label Analyzer
BEFORE: Complex food ingredient labels are hard to interpret for health risks.
AFTER: Built a functional MVP in 6 days using OCR and GPT-4o for nutrition scanning.
6 Days
Sprint to MVP
DIPEX
State Selection
Anomaly-Based IDS
BEFORE: Traditional network security systems lack adaptive anomaly detection baselines.
AFTER: Implemented Deep RL (DQN, PPO, A3C) with Hindsight Experience Replay in PyTorch.
Copyright
Registered (Methodology)
High
Detection Accuracy
EDU
Education & Certifications
B.Tech in Information Technology
Sanjivani College Of Engineering, Kopargaon
2022 - Present | CGPA: 6.7/10
Professional Certifications
- ● [Google] Intro to Generative AI
- ● [Oracle] AI & ML in Java
- ● [NPTEL] Python for Data Science
- ● [Infosys] DSA in Java
JP
Next Move 🇯🇵
Relocating to Osaka, Japan
Targeting AI Engineer Internships (e.g., Futsupa) in manufacturing AI.
Japanese Proficiency:
JLPT N5 CERTIFIED
"
"His approach replaces unnecessary chaos with clear, actionable direction. A true architect of logical strategy."
WHY Institute Assessment
Make Sense Profile
"
"Exceptionally curious and imaginative. Uses high conscientiousness to execute innovative solutions systematically."
"
"Led national-level AI teams at Smart India Hackathon, captained the DIPEX 2025 winning project, and hosted AWS Cloud events for 600+ developers."
MY PROCESS
[ 01 ]
Diagnose Complexity
Take in complex information, analyze patterns, and identify the root cause of the problem rather than treating the surface.
→
[ 02 ]
Architect Logical Strategy
Design a clear, structured architecture built on logical reasoning and data, replacing unnecessary chaos with actionable direction.
→
[ 03 ]
Develop Smart Solutions
Utilize advanced technologies (Agentic AI, Computer Vision, robust APIs) to build the functional MVP or system pipeline.
→
[ 04 ]
Rigorous Testing
Implement thorough testing and monitoring to measure real-world performance and ensure high accuracy.
READY TO BUILD
SOMETHING GREAT?
Maharashtra, India • Open to Projects
Email Me →
nagosejayraj2005@gmail.com