B.Tech CSE (AI & Data Science) — IIIT Kottayam, Class of 2027
Microsoft Certified: Azure AI Engineer Associate
Competitive Programmer — 350+ LeetCode problems solved
Hackathon Builder — Varroc Eureka 3.0 & Smart India Hackathon
Powerlifting Champion — IIIT-K Sports Meet 2026

I build end-to-end AI systems — from computer vision pipelines and real-time backends to full-stack dashboards. My work spans automotive, healthcare, aerospace, and education, with a focus on solutions that are technically rigorous and practically deployable.

Check out my full portfolio: portfolio-akshay-sriram.vercel.app

Varroc Eureka 3.0 Hackathon  |  Python · OpenCV · YOLOv8 · FastAPI · React · NumPy

Real-time ADAS system that scores two-wheeler riding behavior 0–100 with live video analysis.

• 60fps WebSocket video streaming with React + FastAPI dashboard
• Constant-Acceleration Kalman filter on ego-motion-compensated centroids via Lucas-Kanade optical flow
• Calibration-free speed estimation using fleet-shared MPP (m/px = w_real / w_bbox)
• 6 behavioral metrics: hard brake, aggressive acceleration, lane weave, tailgating, sudden stop, helmet non-compliance
• Helmet detection via Otsu blob analysis with 30-frame majority vote + Re-ID by proximity

Next.js 14 · TypeScript · PostgreSQL · Prisma · Tailwind CSS · Nodemailer

Full-stack Cal.com-style scheduling app with production-grade architecture.

• Event type CRUD with unique URL slugs & per-day availability rules with timezone support
• Real-time double-booking prevention via half-open interval overlap checks
• Normalized 6-table PostgreSQL schema with composite indexes for fast slot queries
• Automated email notifications for confirmations, cancellations & 24-hour reminders via secure Cron jobs
• 11 REST API endpoints · Deployed on Vercel

Python · TensorFlow/Keras · Qiskit · NumPy · Pandas

Hybrid quantum-classical deep learning model for medical image classification.

• Combined CNNs with variational quantum circuits using angle encoding with 4 qubits
• 6–8% accuracy improvement over classical baseline → 90% classification accuracy
• Evaluated via precision, recall, F1-score, confusion matrix, and quantum circuit visualizations

Non-Invasive Detection of Diabetes  |  Research — IIT Bombay (Dr. Mahesh Parihar)

• Trained classification models (Logistic Regression, KNN, Decision Trees) on Pima Indians Diabetes Dataset
• Achieved 96% accuracy with ROC-AUC benchmarking

Estimation of Failure of Airplane Engines  |  Student Researcher — DIAT, Pune (Dr. Yogeshwar Singh)

• Built UI dashboard for data preprocessing and fusion workflows
• Deployed ML models for Remaining Useful Life (RUL) estimation on the C-MAPSS dataset

{
  "languages":     ["C", "C++", "Python", "JavaScript", "TypeScript", "Java", "SQL"],
  "ai_ml":         ["TensorFlow", "Keras", "Qiskit", "OpenCV", "YOLOv8", "Scikit-learn"],
  "web":           ["React", "Next.js", "FastAPI", "Node.js", "Tailwind CSS", "Prisma"],
  "databases":     ["PostgreSQL", "SQL"],
  "tools":         ["Git", "Postman", "Docker"],
  "cloud":         ["Microsoft Azure (AI Engineer Associate ✓)"],
  "core_cs":       ["OOPS", "OS", "DBMS", "System Design", "Compiler Design", "Computer Networks"]
}

Varroc Eureka 3.0 Hackathon — Built ADAS scoring system (Problem Statement 3)
Smart India Hackathon 2024 — Participant
Powerlifting — Winner, Intra IIIT-K Sports Meet 2026
Basketball — Winners, School-level & Intra IIIT-K Sports Meet 2023
350+ LeetCode problems solved (DSA)
Microsoft Azure AI Engineer Associate — Certified (Valid: July 2025 – July 2026)

Got a project idea, research opportunity, or just want to connect?

Email — akshaysriram.b@gmail.com