Vignesh N — AI/ML Engineer

// Profile

Turning raw signal into intelligence.

AI/ML engineer focused on the messy edge of the real world — video feeds, dense documents, and live news streams — and the systems that turn them into decisions.

At Shyena Solutions I lead the machine-learning work on an autonomous drone-surveillance platform and build the agentic, GraphRAG core of an enterprise OSINT intelligence system. Alongside that I've shipped production full-stack software at Codegnan, and earlier benchmarked deep-learning architectures for maritime vessel recognition at DRDO's CAIR. I like owning problems end to end: from dataset and annotation, through model training and evaluation, to optimization and deployment.

RoleAI/ML Engineer

Based inBengaluru, India

FocusCV · LLMs · Graphs

DegreeB.E. CSE — DSCE '24

StatusOpen to AI/ML roles

C-01

Computer Vision

Real-time detection & tracking — YOLOv26n, ByteTrack, and ConvNeXt classification, from dataset to deployment.

C-02

LLMs & Agentic AI

Multi-agent orchestration with CrewAI, LangGraph, and Agent Zero over locally-hosted Llama 3.1.

C-03

Retrieval & Graphs

GraphRAG on Neo4j and Qdrant for context-aware retrieval across large OSINT corpora.

C-04

NLP / NER

Domain-adaptive BERT with DAPT, BILOU tagging, and active-learning annotation pipelines.

// Capabilities

Technical stack

The tools I reach for, grouped by where they sit in the pipeline.

01 Languages

PythonSQLC/C++

02 ML & Deep Learning

PyTorchHF TransformersspaCyBERTNERDAPTFine-TuningActive LearningModel Eval

03 LLMs & Agentic AI

LLMsMulti-Agent AICrewAILangGraphAgent ZeroRAGGraphRAGPrompt EngineeringOllamaLlama 3.1

04 Computer Vision

YOLOv26nConvNeXtByteTrackObject DetectionImage ClassificationObject TrackingOpenCV

05 Graphs & Databases

Neo4jQdrantPostgreSQLRedis / Valkey

06 Data & MLOps

ScrapyTrafilaturaDoccanoLabel StudioPaddleOCRTesseractPyMuPDFDoclingDockerFastAPIGitREST APIs

// Fieldwork

Experience

AI/ML Engineer InternNov 2025 — Present

Shyena Solutions · Bengaluru, India

Designed and fine-tuned domain-adaptive BERT models for maritime Named Entity Recognition — running DAPT, migrating annotations from BIO to BILOU, and automating labeling with spaCy across multiple active-learning iterations.
Engineered the GraphRAG, knowledge-graph, and multi-agent AI components of an enterprise OSINT platform using LangGraph, CrewAI, and Agent Zero, integrating Ollama-hosted Llama 3.1 with Neo4j, Docker, and FastAPI.
Built the OSINT data-ingestion pipeline with Scrapy, Trafilatura, and PostgreSQL — de-duplication, domain filtering, and structured logging feeding downstream NLP, GraphRAG, and analysis systems.

BERTGraphRAGCrewAILangGraphNeo4jFastAPI

Software Development InternJul 2024 — Present

Codegnan · India

Built production full-stack features across the MERN stack and Next.js — functional React components, responsive Tailwind CSS interfaces, and Express.js / MongoDB backends.
Designed and tested RESTful APIs with Postman, and shipped projects to production on Netlify and Vercel with streamlined CI/CD workflows.
Managed version control with Git to maintain clean, production-ready code — the engineering foundation I now bring to ML systems and MLOps.

ReactNext.jsExpress.jsMongoDBREST APIsCI/CD

Deep Learning Research InternFeb 2024 — May 2024

Centre for Artificial Intelligence & Robotics (CAIR), DRDO · Bengaluru, India

Conducted a comparative study of MobileNetV2, Xception, VGG-16, and Vision Transformers for ship image classification, benchmarking accuracy and computational efficiency.

Vision TransformersCNNsImage ClassificationBenchmarking

// Operations

Selected projects

Production-grade AI systems built end to end — each with a live look at what it does.

RECCAM-01 · DETECT+TRACK24 FPS

OP-01 · Computer VisionFlagship

Autonomous Drone Detection, Tracking & Classification

YOLOv26n · ConvNeXt · ByteTrack · Anti-UAV · PTZ

Leading ML for a production-grade autonomous drone-surveillance platform — owning the full pipeline: detection, tracking, classification, training, evaluation, optimization, deployment.
Fine-tuning YOLOv26n for real-time detection on the Anti-UAV dataset, with ByteTrack multi-object tracking and a ConvNeXt classifier for drone family/model recognition.
Designing dataset prep, augmentation, hyperparameter optimization, benchmarking, inference optimization, and PTZ camera integration.

Role: ML LeadReal-timeEdge deploy

LIVEORCHESTRATOR · 5 AGENTSGraphRAG

OP-02 · Agentic AI

OSINT Multi-Agent Intelligence Platform

CrewAI · LangGraph · Agent Zero · Ollama · Neo4j · Valkey

Developed the AI backbone of a production-grade OSINT platform using CrewAI, LangGraph, and Agent Zero to orchestrate retrieval, web search, reasoning, memory, and summarization.
Implemented a GraphRAG architecture over Neo4j, Valkey/Redis, and Ollama-hosted LLMs for context-aware retrieval and conversational intelligence across large-scale data.

Multi-agentOn-prem LLMsConversational

PROCPARSE · CHUNK · EMBEDQdrant

OP-03 · Document AI

OSINT Multi-Modal Document Intelligence Pipeline

PyMuPDF · Docling · PaddleOCR · Tesseract · Qdrant

Developed the document-ingestion pipeline for the OSINT platform, extracting text, tables, and images from complex technical PDFs.
Built a semantic indexing pipeline on Qdrant for high-accuracy retrieval and RAG over multimodal intelligence documents.

Multi-modalOCRVector search

// Credentials

Education & recognition

Education

B.E. — Computer Science & Engineering

Dayananda Sagar College of Engineering

Graduated June 2024GPA 7.72 / 10

Recognition

2ND

hAIothon — Runner-up

2nd place at an inter-college Artificial Intelligence hackathon.

BEST

Best Project Award

Awarded at the college Project Open Day.

// Dossier

Résumé

The full record — viewable here, or save a clean PDF straight from your browser.

VIGNESH N

8951288231 | vigneshn6236@gmail.com | linkedin.com/in/vigneshn6236 | github.com/vignesh6236 | Portfolio

Professional Summary

Artificial Intelligence and Machine Learning engineer specializing in Natural Language Processing, Computer Vision, LLMs, Multi-Agent AI Frameworks, Knowledge Graphs, and Retrieval-Augmented Generation (RAG). Experienced in building production-grade AI systems spanning data collection, annotation, model fine-tuning, evaluation, and deployment across computer vision, NER, GraphRAG, autonomous drone surveillance, and OSINT intelligence platforms.

Technical Skills

Languages: Python, SQL, C/C++
ML & Deep Learning: PyTorch, Hugging Face Transformers, spaCy, BERT, Named Entity Recognition (NER), Domain-Adaptive Pretraining (DAPT), Fine-Tuning, Active Learning, Model Evaluation
LLMs & Agentic AI: LLMs, Multi-Agent AI, Agentic AI, CrewAI, LangGraph, Agent Zero, Retrieval-Augmented Generation (RAG), GraphRAG, Prompt Engineering, Ollama, Llama 3.1
Computer Vision: YOLOv26n, ConvNeXt, ByteTrack, Object Detection, Image Classification, Object Tracking, OpenCV
Graphs & Databases: Neo4j, Qdrant (Vector DB), PostgreSQL, Redis / Valkey
Data & MLOps: Web Scraping (Scrapy, Trafilatura), Annotation (Doccano, Label Studio), OCR (PaddleOCR, Tesseract), PyMuPDF, Docling, Docker, FastAPI, Git, REST APIs

Experience

Shyena Solutions Nov 2025 – Present

Intern Bengaluru, India

Designed and fine-tuned domain-adaptive BERT models for maritime Named Entity Recognition, performing DAPT, migrating annotations from BIO to BILOU, and automating labeling using spaCy through multiple active-learning iterations.
Engineered the GraphRAG, knowledge graph, and multi-agent AI components of an enterprise OSINT platform using LangGraph, CrewAI, and Agent Zero, integrating Ollama-hosted Llama 3.1 with Neo4j, Docker, and FastAPI.
Built the OSINT data-ingestion pipeline using Scrapy, Trafilatura, and PostgreSQL, implementing de-duplication, domain filtering, and structured logging for downstream NLP, GraphRAG, and intelligence-analysis systems.

Centre for Artificial Intelligence and Robotics (CAIR), DRDO Feb 2024 – May 2024

Deep Learning Research Intern Bengaluru, India

Conducted a comparative study of MobileNetV2, Xception, VGG-16, and Vision Transformers for ship image classification, benchmarking accuracy and computational efficiency.

Key Projects

Autonomous Drone Detection, Tracking & Classification Platform YOLOv26n, ConvNeXt, ByteTrack, Anti-UAV

Leading the machine learning development of a production-grade autonomous drone surveillance platform, owning the end-to-end AI pipeline covering detection, tracking, classification, model training, evaluation, optimization, and deployment.
Fine-tuning YOLOv26n for real-time drone detection on the Anti-UAV dataset, integrating ByteTrack for multi-object tracking, and developing a ConvNeXt-based classifier for drone family/model recognition.
Designing the complete computer vision pipeline including dataset preparation, augmentation, hyperparameter optimization, benchmarking, inference optimization, and PTZ camera integration.

OSINT Multi-Agent Intelligence Platform CrewAI, LangGraph, Agent Zero, Ollama, Neo4j, Valkey

Developed the AI backbone of a production-grade OSINT platform using CrewAI, LangGraph, and Agent Zero to orchestrate retrieval, web search, reasoning, memory, and intelligence summarization.
Implemented a GraphRAG architecture integrating Neo4j, Valkey/Redis, and Ollama-hosted LLMs for context-aware retrieval and conversational intelligence across large-scale OSINT data.

OSINT Multi-Modal Document Intelligence Pipeline PyMuPDF, Docling, PaddleOCR, Tesseract, Qdrant

Developed the document-ingestion pipeline for the OSINT platform, extracting text, tables, and images from complex technical PDFs.
Built a semantic indexing pipeline using Qdrant to enable high-accuracy retrieval and RAG over multimodal intelligence documents.

// Contact

Let's build something that thinks.

Open to AI/ML engineering roles and collaborations in computer vision, LLM systems, and applied intelligence. The fastest way to reach me is email.

Say hello ↗

Emailvigneshn6236@gmail.com Phone+91 89512 88231 LinkedIn/in/vigneshn6236 GitHub/vignesh6236 Portfoliovignesh-n.netlify.app