Atharva Sarde | Portfolio

About Me

7+ Projects Built

6+ Certifications

Full Stack Developer

AI/ML Researcher

atharva_bio.json

const atharva = {
  role: "Computer Science Engineer (AI & ML)",
  university: "Universal AI University",
  location: "Pune / Pimpri-Chinchwad",
  currently: "AI Intern @ RAMS Digital",
  passions: ["Deep Learning", "Edge AI", "Agentic AI", "3D Design"]
};

// Bio
I am an innovator at heart. Currently pursuing my B.Tech and interning as an AI Intern at RAMS Digital, I live at the intersection of AI Innovation and Creative Design. My journey involves building complex Dual-Stage Deep Learning Frameworks for medical diagnosis, architecting autonomous systems for hospitals, and taking YOLO models from PyTorch through ONNX and MLIR to INT8-quantized edge deployments.

Beyond code, I led the AI Odyssey 1.0 hackathon and actively contribute to the tech community. I combine technical rigor with a designer's eye to create meaningful digital solutions.

Education Journey

2024 - 2028

Universal AI University

B.Tech in CSE (AI & ML)

Specializing in Deep Learning architectures, Neural Networks, and Autonomous Agents. Active participant in technical symposiums and hackathons.

2022 - 2024

Alard Public School

Higher Secondary Certificate

Graduated with First Division Distinction. Built a strong foundation in Mathematics, Physics, and Computer Science fundamentals.

Professional Experience

My contributions to the industry and leadership roles.

Artificial Intelligence Intern

RAMS Digital, Pune Division

Apr 2026 - Jul 2026

Built an end-to-end YOLO computer vision pipeline for real-time rack/warehouse defect detection, from dataset curation and class rebalancing to production deployment.
Took models from PyTorch (FP32) → ONNX → MLIR → INT8-quantized .cvimodel and deployed on MaixCAM / CVITEK NPU edge hardware for real-time on-device inference with no cloud dependency.
Built a real-time PPE compliance detection system and a smart MHE safety camera with a face-authentication interlock gating the equipment's brake circuit.
Contributed to SALT, an AI-powered CAD system generating consultant-grade DXF/SVG drawing packages for water infrastructure, building the structural synthesis engine and export pipeline.

Graphic Designer

Sparge Chem Private Limited

Sept 2022 - Apr 2026

Created 50+ digital marketing creatives boosting client engagement by 40%.
Integrated HR and Accounting visuals into B2B workflows.
Leveraged design analytics to align visuals with business objectives.

Creative Lead & Organizer

AI Odyssey 1.0 @ Universal AI University

Mar 2025 - Oct 2025

Spearheaded branding for hackathon with 200+ participants.
Produced promotional materials improving visibility by 60%.
Received Certificate of Appreciation for outstanding leadership.

Featured Projects

A selection of my recent R&D work and Full Stack applications.

Research

MediScopic-BC

Dual-stage deep learning framework for breast cancer histopathology, reaching 84.84% accuracy. Introspection layer using Entropy & Embedding Distance flags ~75% of uncertain predictions for human review.

Python PyTorch ResNet50

Code

Hackathon Win

HospAgent

Digital health record system for migrant workers. AI-driven patient surge prediction improving capacity planning by 35%.

React Node.js Agentic AI

Code

Deployed

SpargeChem

Production CRO/CDMO platform for a custom-synthesis chemical manufacturer, with an RDKit/3Dmol-powered molecule search and catalog, admin automation, and a PostgreSQL backend on a self-hosted VPS.

Next.js RDKit PostgreSQL

Live Site

Edge AI

Edge AI Defect & PPE Detection

YOLO-based rack/warehouse defect detection and real-time PPE compliance, quantized to INT8 and deployed on CVITEK NPU / MaixCAM edge hardware with no cloud dependency.

Ultralytics YOLO ONNX INT8 Quantization

Code

CAD Automation

SALT

AI-powered system generating consultant-grade multi-sheet DXF/SVG drawing packages for water infrastructure, with a precedent library from 38 real WTP drawings across 9 projects.

Python FastAPI DXF/SVG

Code

View all projects on GitHub

Technical Proficiency

Languages

Python JavaScript C++ SQL

AI & ML

PyTorch TensorFlow OpenCV CNNs Transfer Learning Grad-CAM Generative AI

Edge AI

Ultralytics YOLO INT8 Quantization ONNX MLIR Model Calibration CVITEK NPU / MaixCAM

Web & DB

React Node.js FastAPI REST APIs PostgreSQL Supabase MQTT

Tools

Docker AWS Git Linux Figma

Ready to Collaborate?

Whether you have a question, a project idea, or just want to discuss the latest in AI, I'm always open to connecting.

Email Me Connect on LinkedIn