Darshan Joshi
Darshan Joshi
Software Engineer
Available for Opportunities

Darshan
Joshi

Building AI systems, full stack applications, and backend services. Research Assistant at Lawrence Tech — 2 publications, MS in CS (GPA 3.35/4.0). Open to Software Engineer, AI, and Full Stack roles.

View ProjectsContact MeRésumé
0
Master's GPA
0
Publications
0
Projects
Scroll
01 —

About Me

I'm a Software Engineer building AI systems, full stack applications, and backend services. Currently a Research Assistant at Lawrence Technological University, working on real-time computer vision systems and deep learning research.

My work spans computer vision (YOLOv8, PyTorch), full stack development (FastAPI, React, SQLite), NLP pipelines, database design (MySQL, 3NF), Docker deployment, and cybersecurity research.

I have 2 peer-reviewed publications in cybersecurity and deep learning, 9 engineering projects, and an MS in Computer Science (GPA 3.75/4.0) from Lawrence Technological University.

PythonPyTorchYOLOv8xYOLOv5Grad-CAMOpenCVScikit-LearnXGBoostProphetFastAPINLPLLMREST APIsPydanticSQLitePandasNumPyMatplotlibSeabornReportLabpython-docxGDBC / C++x86 ASMMySQLSQLDockerReactViteGitGoogle ColabRoboflowYAMLJupyter
LocationSouthfield, MI, USA
StatusOpen to Opportunities
DegreeM.S. Computer Science
GPA3.35 / 4.0
Publications2 Research Papers
Emailjoshidarshan193@gmail.com
Download Résumé
02 —

Experience

Research Assistant

Lawrence Technological University · Southfield, MI
CurrentDec 2024 — Present
  • ›Led end-to-end development of an AI-based classroom behavior detection system using YOLOv8 with measurable Precision, Recall, and mAP performance metrics.
  • ›Designed and curated structured multi-class datasets (11+ behavioral classes), enforcing annotation quality under limited hardware resources.
  • ›Integrated Grad-CAM explainability techniques to enhance model interpretability and support responsible AI practices.
  • ›Conducted iterative hyperparameter optimization and documented reproducible YAML-based training workflows.
  • ›Collaborated with faculty advisors to refine experimental scope and presented findings in structured research updates.
  • ›Independently troubleshot training instability, environment configuration issues, and dataset inconsistencies.
  • ›Maintained organized documentation of experiments, configuration files, and dataset revisions for long-term sustainability.
03 —

Selected Projects

01Featuredaibackend

Resume Intelligence Engine

Full-Stack AI Job Application System · FastAPI + LLM + ATS

A production-grade AI job application system with 25+ Python modules, a web dashboard, LLM-powered extraction, ATS scoring, recruiter outreach automation, and an adaptive learning engine. Generates tailored resumes and cover letters from job descriptions with 87%+ suitability accuracy.

FastAPIPythonLLMSQLiteNLPJinja2+4
25+
Python Modules
87%+
ATS Suitability
5
Output Formats
View Project →
2026 – Present
02
ai

Classroom Behavior Detection System

YOLOv8x · 74.85% mAP@0.5 · Real-Time Inference

Production-grade classroom behavior detection system trained on the SCB-05 dataset using YOLOv8x on Google Colab Pro (A100 GPU). Achieves 74.85% mAP@0.5 overall and 93.5% detection accuracy on sleeping behavior across 11+ behavioral classes. Includes tiled inference pipeline for live CCTV/RTSP feeds.

YOLOv8xPyTorchGoogle Colab+5
View →
03
security

Buffer Overflow Attack Lab

Cybersecurity Research · Low-Level Systems

Analyzed buffer overflow vulnerabilities in 32-bit environments — stack smashing, heap overflow, format string attacks. Evaluated ASLR and stack canary countermeasures using GDB.

CGDBx86 ASM+3
View →
04
ai

Formal Language Typing Game

Theory of Computation · Tkinter · 9 Language Patterns

A Theory of Computation coursework project — an interactive Tkinter game that challenges players to type strings matching 9 formal language patterns in 45 seconds. Implements DFA (regex), PDA (stack-based balanced parentheses), and string reversal (palindrome) validators from scratch in pure Python.

PythonTkinterRegex (re)+3
View →
05
dbbackend

Global Harvest Imports

Enterprise Relational Database · 3NF · MySQL · 7 Tables

End-to-end relational database design for a multi-department international agricultural import company. Deliverables include a fully normalised 3NF MySQL schema (7 tables, 5 analytical queries), seed data, ERD documentation, and a DFD-driven design process.

MySQLSQLERD+3
View →
06
aibackend

Indoor Scene Recognition

YOLOv5 · REST API · Web App · Docker

A deployable indoor scene recognition system using a fine-tuned YOLOv5s model (14.4MB), served via both a REST API (restapi.py) and an interactive web application (webapp.py) built with Flask/Jinja2. Fully containerized with Docker for one-command deployment.

YOLOv5PyTorchPython+5
View →
07
aidb

Twitter Analytics & Engagement Prediction

Social Network Mining · XGBoost · Prophet · PyTorch · R²=0.852

A complete social network mining and engagement prediction pipeline built on a cleaned Twitter/X timeline dataset. Combines sentiment analysis, temporal behavioral analysis, and four predictive models (XGBoost, Linear Regression, Prophet time-series, PyTorch DL) achieving R²=0.852 on engagement forecasting.

PythonPandasXGBoost+7
View →
08
aibackend

AI Requirements Generator

Software Engineering · FastAPI + React + Vite · SRS Automation

A full-stack Software Engineering coursework project that automates requirements gathering using AI. FastAPI backend exposes a /generate-requirements endpoint; React/Vite frontend lets users input a system description and instantly generates SRS-ready functional requirements.

FastAPIPythonReact+4
View →
09
ai

Advanced Sorting Algorithms Benchmark

TimSort · Radix Sort · Bucket Sort · Python · Performance Analysis

A Machine Learning / Algorithms coursework project implementing and benchmarking three advanced sorting algorithms (TimSort, Radix Sort, Bucket Sort) across three dataset types (random, nearly sorted, reversed) at sizes from 500 to 100,000 elements. Produces performance comparison visualizations.

PythonNumPyMatplotlib+4
View →
All Projects →
04 —

Latest Research

Published2025Deep Learning / Seismology

Earthquake Prediction and Synthetic Seismogram Generation Using Hybrid CNN-LSTM Model

This paper presents a hybrid CNN-LSTM deep learning architecture for earthquake prediction and the generation of synthetic seismograms. The model leverages convolutional layers for spatial feature extraction from seismic waveform data and LSTM layers for temporal sequence modeling, enabling accurate magnitude prediction and realistic synthetic seismogram synthesis for data augmentation and simulation purposes.

CNNLSTMEarthquake PredictionSeismogramDeep Learning
American Journal of Civil Engineering
Read Paper
All Publications
Design and Analysis of Cyber Security Infrastructure in Large Enterprises and Organisations
IJARESM · 2023
Classroom Behavior Detection Using YOLOv8 and Explainable AI
IJSRST · 2025
Earthquake Prediction and Synthetic Seismogram Generation Using Hybrid CNN-LSTM Model
American Journal of Civil Engineering · 2025
View All Research →
Currently building:Resume Intelligence Engine
Open to Work

Let's build something
remarkable.

Seeking Software Engineer / AI Engineer / Full Stack roles. 2 publications, MS CS 3.35 GPA. Southfield, MI — open to remote.

Get in TouchView Research