sajjad ahmed shaaz

Projects

Things I've built

A mix of research prototypes, real products, and competition systems.

Technical co-founder. Restaurants digitise their menus; diners get a per-restaurant AI chatbot delivering personalised dish recommendations via RAG — an open-source LLM (Groq API) interprets user intent and ranks dishes by cosine similarity over per-dish ChromaDB vector embeddings.

Next.jsDjangoPostgreSQLChromaDBGroq APIRAG

HawkEye

In Development

Smart drone communication & GPS-resilient navigation

Leading end-to-end development at Aarish Technologies. CNN-based anti-jamming positioning: IMU inputs → velocity prediction (vx, vy, vz) → dead-reckoning GPS estimation. Full Android MAVLink stack, dual-SIM cellular health monitoring, adaptive video streaming, and a real-time browser-based ground control dashboard.

AndroidMAVLinkPyTorchReactWebSocketsPixhawk

NIDAR

Finalistread more →

Autonomous multi-UAV precision agriculture system

5th nationally

Dual-UAV pipeline built under JUMTC. Scan drone runs YOLO-based crop health inference onboard; geo-tagged fertilization targets are relayed over MAVLink telemetry to a spray drone executing closed-loop waypoint spraying — fully autonomous, end-to-end.

YOLOOpenCVMAVLinkArduPilotPythonROS

Obseract

Finalists — IRC

Mars rover — International Rover Challenge

JUMTC rover project. Built YOLO-based object detection and terrain CV pipeline for field task execution. Implemented autonomous arm control using Stable-Baselines3 PPO with a custom OpenAI Gym environment and shaped reward function for manipulator task completion.

YOLOStable-Baselines3PPOOpenCVPythonROS

BusKaro

Prototyperead more →

Real-time public transit app for Kolkata

Mobile app addressing Kolkata's untracked bus network: live GPS tracking with ETA, digital smart ticketing, AI-assisted route planning, and crowd monitoring. Validated with 255-response user survey and on-ground operator interviews. Gamified retention layer with Route Rally daily game.

React NativeFirebaseGoogle Maps APINode.js

CNN Model Compression

3rd — Srijanread more →

105× model size reduction with 7.6× latency speedup

42 MB → 0.4 MB · 68 ms → 8.9 ms

Compressed a 42 MB image classification CNN to 0.4 MB using structured pruning, INT8 post-training quantization, and knowledge distillation — reducing inference latency from 68 ms to 8.9 ms with only a 3–5% accuracy drop. 3rd place at Srijan OpenAImer.

PyTorchStructured PruningINT8 QuantizationKnowledge Distillation

CheatCode

Interactive coding learning platform

LeetCode-style platform where students spin up isolated Docker containers with real-time Linux terminals over WebSockets. Adaptive learning paths adjust to user performance with no shared session state between containers.

ReactNode.jsDockerWebSockets

Vendor Management System

Multi-tenant B2B order automation platform

Multi-tenant B2B platform for vendor and order automation: catalog management, automated purchase orders, invoicing, and returns workflows. Razorpay payment disbursal; Docker-containerised for tenant isolation.

ReactDjangoPostgreSQLDockerRazorpay

Statera 2.0

Live

Event management platform — JUMTC

Full-stack club event platform: Clerk auth with multi-device session handling, Cashfree payments, Prisma ORM on PostgreSQL, Vercel Blob uploads, and a 3D interactive venue map built with React Three Fiber.

Next.jsPrismaClerkCashfreeReact Three Fiber