I'm Yugandhar, a software engineer specializing in computer vision, deep learning, and full-stack product engineering. My work sits at the intersection of research and production — I don't just prototype models, I engineer them into systems that run in real time, at scale, and under real-world constraints.
My core engineering focus spans:
- 🧠 AI/ML Systems Engineering — building identity-recognition pipelines using ArcFace embeddings, FAISS + HNSW vector indexing, and approximate nearest-neighbor search for sub-second retrieval at scale.
- ⚙️ Full-Stack Product Engineering — designing and shipping real-time dashboards, monitoring systems, and internal tools with Django, FastAPI, and modern JS tooling.
- 🔬 Applied Research Engineering — translating academic techniques (CLAHE, edge detection, CNN architectures) into deployable diagnostic and vision pipelines.
- 🛡️ Systems & Security Mindset — network scanning, vulnerability assessment, and infrastructure automation (Jenkins, CI/CD).
I approach engineering with a product mindset: performance, security, and scalability aren't afterthoughts — they're first-class requirements from day one.
🎯 Open To:
roles: [Software Engineer, AI/ML Engineer, Backend Engineer, Computer Vision Engineer]
collaboration: [Open Source AI/ML, Deep Learning Pipelines, Hackathons, Drone Tech & IoT]
availability: Full-time / Internship / Open-Source CollaborationLanguages
Frontend
Backend & Databases
Cloud, DevOps & Tooling
| Domain | Proficiency | Details |
|---|---|---|
| Computer Vision | ⭐⭐⭐⭐⭐ | ArcFace embeddings, CNN architectures, YOLO real-time detection, DBSCAN & Cosine Similarity clustering |
| Vector Search & ANN | ⭐⭐⭐⭐ | FAISS, HNSW indexing, L2-normalization, sub-second identity retrieval at scale |
| Deep Learning | ⭐⭐⭐⭐ | TensorFlow/Keras, PyTorch, custom CNNs for feature extraction & classification |
| Medical Imaging AI | ⭐⭐⭐⭐ | CLAHE preprocessing, edge detection (Canny/Sobel), diagnostic model optimization (DFUC 2021) |
| MLOps & Deployment | ⭐⭐⭐ | Local-to-cloud model scaling, Streamlit deployment, production-ready inference pipelines |
| Data Science | ⭐⭐⭐⭐ | NumPy, Pandas, scikit-learn, Matplotlib, Plotly for analysis & visualization |
🔐 FaceVault — Privacy-Preserving Facial Clustering System
Privacy-first, dynamic facial clustering system that leverages ArcFace embeddings, FAISS-HNSW vector indexing, and e-Differential Privacy federated sync to deliver secure, scalable identity clustering without compromising user privacy.
| Attribute | Detail |
|---|---|
| Stack | Python, ArcFace, FAISS, HNSW, Streamlit |
| Scale | Sub-second ANN retrieval across large embedding sets |
| Performance | Optimized vector search via HNSW graph indexing |
| Security | e-Differential Privacy for federated identity sync |
| Impact | Enables privacy-preserving identity management at scale |
| Repository | FaceVault |
Built to solve the core tension between facial recognition utility and user privacy — combining state-of-the-art embedding models with differential privacy guarantees, wrapped in an accessible Streamlit interface for real-world usability.
🎯 YOLO-Live-Ident — Real-Time Multi-Threaded Identification Pipeline
A real-time, multi-threaded facial identification and tracking pipeline combining YOLO object detection with deep feature embeddings for concurrent, low-latency identity tracking.
| Attribute | Detail |
|---|---|
| Stack | Python, YOLO, Deep Feature Embeddings, Multi-threading |
| Scale | Real-time multi-stream video processing |
| Performance | Threaded architecture for concurrent detection & tracking |
| Security | Local inference — no external data transmission |
| Impact | Foundation for live surveillance & identity monitoring systems |
| Repository | YOLO-Live-Ident |
Engineered to close the gap between offline batch identification and real-time production requirements, using threading to keep detection and embedding extraction decoupled and performant.
🧠 Person-Identification-Using-CNN — Custom Deep Learning Framework
End-to-end deep learning framework for person identification and tracking, built on custom Convolutional Neural Networks with TensorFlow/Keras for high-accuracy feature extraction.
| Attribute | Detail |
|---|---|
| Stack | Python, TensorFlow, Keras, Custom CNN Architectures |
| Scale | Trained on multi-class identity datasets |
| Performance | High-accuracy feature extraction via tuned CNN layers |
| Security | Local model training & inference |
| Impact | Reusable framework for identity classification tasks |
| Repository | Person-Identification-Using-CNN |
Designed from the ground up rather than relying on pretrained backbones, giving full control over the feature extraction pipeline and enabling deep architectural experimentation.
🛡️ NetShield-Vulnerability-Scanner — Automated Network Security Auditor
A Python/Flask-based local network scanner that discovers active hosts, evaluates vulnerabilities using CVSS scoring, and auto-generates professional PDF audit reports.
| Attribute | Detail |
|---|---|
| Stack | Python, Flask, CVSS Scoring, PDF Reporting |
| Scale | Full local subnet host discovery & scanning |
| Performance | Automated end-to-end scan-to-report pipeline |
| Security | CVSS-based vulnerability severity classification |
| Impact | Streamlines network security auditing for small networks |
| Repository | NetShield-Vulnerability-Scanner |
Built to bring enterprise-style vulnerability auditing practices into a lightweight, self-hosted tool suitable for local network administrators.
⚙️ Jenkins-Project — CI/CD Automation Pipeline
A CI/CD automation project demonstrating build, test, and deployment pipeline orchestration using Jenkins.
| Attribute | Detail |
|---|---|
| Stack | Jenkins, Git, CI/CD Pipelines |
| Scale | Automated multi-stage build pipeline |
| Performance | Reduced manual deployment overhead |
| Security | Controlled, repeatable deployment process |
| Impact | Demonstrates DevOps automation proficiency |
| Repository | Jenkins-Project |
Reflects a DevOps-first mindset — treating build and deployment automation as a core engineering discipline rather than an operational afterthought.
Internship
Enhanced Django-based dashboards for server room metric visualization and anomaly detection, supporting real-time infrastructure monitoring for critical meteorological systems.
Scope of Work:
- Built and refined real-time monitoring dashboards for server room environmental metrics
- Implemented anomaly detection logic to flag infrastructure irregularities
- Optimized backend data pipelines for low-latency visualization
Django Python Data Visualization Anomaly Detection Backend Engineering
Independent Research — DFUC 2021 Challenge
Optimized image preprocessing pipelines for diagnostic deep learning models, with a focus on CLAHE contrast enhancement and edge detection for improved diagnostic accuracy.
Scope of Work:
- Implemented CLAHE-based preprocessing for medical image enhancement
- Applied Canny/Sobel edge detection for feature refinement
- Benchmarked preprocessing impact on downstream model performance
Python OpenCV Medical Imaging CLAHE Deep Learning
Technical Leadership
Managed technical teams and led club initiatives, coordinating projects, technical events, and cross-team collaboration.
Scope of Work:
- Led technical team operations and project coordination
- Mentored peers on full-stack and AI/ML project development
- Organized technical events and hackathon preparation
Team Leadership Project Management Technical Mentorship
| Recognition | Details |
|---|---|
| 🥉 2nd Runner-Up | National Hackathon at IIT Ropar |
| 🏅 Top 30 Finalist | National Hackathon out of 9,000+ participants |
| 🎓 NITTTR & ISRO Certified | Drone Technology & IoT Systems |
| 📜 Kusum Lata Award | Sanskrit Language Proficiency |
| 🎸 1st Place | Bhagavad Gita Talent Search (Multiple Editions) |
| 🏀 Best Basketball Player | School-Level Recognition |
Section coming soon — verified certificates to be added.
learning:
- Hierarchical Navigable Small Worlds (HNSW) & FAISS for ANN search
- Advanced Vector Embeddings & high-dimensional similarity search
- Ethical Hacking (Kali Linux, Unix/Linux security)
building:
- Scalable computer vision systems (face clustering, real-time identification)
- Medical AI diagnostic preprocessing pipelines
- Real-time infrastructure monitoring dashboards
exploring:
- Isolated LAN configurations & peer-to-peer communication protocols
- MLOps deployment of local models into cloud-native production
- Drone-guided information retrieval systems
open_to:
- Software Engineering & AI/ML Engineering roles
- Open-source deep learning & identity management projects
- Hackathons & high-impact technical collaborations
"Engineering is the art of turning research into systems that work — reliably, at scale, and in production."