Basharameez

SHAIK RAMEEZ BASHA

AI SYSTEMS ENGINEER

AI/ML · FULL-STACK · PRODUCTION SOFTWARE

I build intelligent software systems — from models and data pipelines to APIs, interfaces, infrastructure, and production workflows.

flowchart LR
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef emerald fill:#10B981,stroke:#059669,color:#FFFFFF,stroke-width:2px;
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef amber fill:#F59E0B,stroke:#D97706,color:#FFFFFF,stroke-width:2px;
    classDef indigo fill:#6366F1,stroke:#4F46E5,color:#FFFFFF,stroke-width:2px;

    A[AI Systems]:::indigo --> B[Data]:::cyan
    B --> C[Models]:::violet
    C --> D[APIs]:::blue
    D --> E[Infrastructure]:::emerald
    E --> F[Products]:::amber

HOME — AI ENGINEERING SYSTEM

FEATURED SYSTEMS TILES

_{AI Recruitment OS}

_{Industrial AI Telemetry}

_{ML Robustness Benchmark}

_{Biomedical Vision Pipeline}

PyTorch & Computer Vision
NLP & LLM Applications
Explainable AI (SHAP, Integrated Gradients)
ML Robustness & ECE Calibration

Full-Stack SaaS (Next.js 15, React 19, TypeScript)
Python, FastAPI, Express.js & Node.js
Async Queues (BullMQ, Redis Workers)
PostgreSQL, Drizzle ORM, Supabase, MongoDB

Signal Processing (4,096-line FFT, Vibration Spectrum)
Industrial Intelligence (ISO 10816, BPFO / BPFI)
Test-Driven Rigor (170 Vitest, 39 PyTest)
Docker & Containerized Workflows

`170 / 170`

_{Vitest Tests Passed}

`39 / 39`

_{PyTest Suite Passed}

`11 / 11`

_{API Verification Tests}

`01`

_{IEEE Publication}

↑ HOME DOCK

02 — SELECTED BUILDS

`01` — APTIVUE

AI-NATIVE RECRUITMENT INFRASTRUCTURE |

Aptivue (AptiHire AI / TalentOS) is an AI-powered recruitment evaluation platform designed around resume intelligence, candidate evaluation, semantic matching, hiring workflows, and analytics.

`170 / 170`

_{Vitest Unit & Integration Tests Passed}

`38`

_{Test Files Executed}

flowchart LR
    classDef indigo fill:#6366F1,stroke:#4F46E5,color:#FFFFFF,stroke-width:2px;
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef emerald fill:#10B981,stroke:#059669,color:#FFFFFF,stroke-width:2px;
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;

    A[Candidate Resume]:::indigo --> B[Resume Parser]:::blue
    B --> C[(PostgreSQL DB)]:::blue
    B --> D[Redis Queue Cache]:::emerald
    D --> E[BullMQ Worker]:::emerald
    E --> F[Gemini LLM]:::violet
    F --> G[AI Evaluation]:::indigo
    G --> H[Semantic Matching]:::cyan
    H --> I[Hiring Workflow]:::cyan
    I --> J[Analytics Dashboard]:::cyan

Tech Stack: Next.js 15 · React · TypeScript · Gemini · PostgreSQL · Supabase · Drizzle ORM · Redis · BullMQ · Vitest
Repository Link: AptiHire-AI GitHub Repository

`02` — ROTORDYN

INDUSTRIAL VIBRATION INTELLIGENCE |

A production-oriented SaaS platform for analyzing machine vibration telemetry and transforming raw sensor data into actionable bearing diagnostics, spectral peak detection, and ISO 10816 fault severity classification.

`4,096-Line`

_{FFT Resolution}

`RMS`

_{Vibration Velocity}

`BPFO / BPFI`

_{Defect Detection}

`ISO 10816`

_{Standard Compliance}

flowchart LR
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;
    classDef emerald fill:#10B981,stroke:#059669,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef dark fill:#1E293B,stroke:#334155,color:#F8FAFC,stroke-width:1px;

    A[CSV / Excel / Sensor Data]:::dark --> B[FastAPI Ingestion]:::blue
    B --> C[Signal Preprocessing]:::cyan
    C --> D[4096-Line FFT Analysis]:::cyan
    D --> E[BPFO / BPFI Detection]:::emerald
    E --> F[RMS + ISO 10816 Severity]:::emerald
    F --> G[Diagnostic Report Dashboard]:::dark

Tech Stack: Python · FastAPI · React · PostgreSQL · Pandas · Plotly.js · FFT · ISO 10816
Capabilities: CSV / Excel ingestion · Signal preprocessing · 4,096-line FFT · RMS velocity · BPFO/BPFI defect frequency · Machine-health severity classification · Interactive visualization · AI-assisted reporting
Repository Link: RotorDyn Enterprise Repository

`03` — BIOROBUST

ML ROBUSTNESS & COMPUTER VISION |

A benchmarking framework for evaluating computer vision model behavior under controlled image degradation, distribution shifts, and Expected Calibration Error (ECE) analysis.

`7,180`

_{Test Images (PathMNIST)}

`35`

_{Perturbation Conditions}

`73.66%`

_{Clean Accuracy}

`0.0782`

_{Clean ECE Calibration}

`39 / 39`

_{PyTest Passed}

flowchart TD
    classDef amber fill:#F59E0B,stroke:#D97706,color:#FFFFFF,stroke-width:2px;
    classDef rose fill:#F43F5E,stroke:#E11D48,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef dark fill:#1E293B,stroke:#334155,color:#F8FAFC,stroke-width:1px;

    A[PathMNIST — 7,180 Images]:::dark --> B[ResNet-18 Baseline]:::blue
    B --> C[Controlled Perturbations]:::amber

    C --> D[Blur Corruptions]:::rose
    C --> E[Resolution Shifts]:::rose
    C --> F[Additive Noise]:::amber
    C --> G[Contrast / Other Conditions]:::amber

    D & E & F & G --> H[Accuracy / F1 / ECE Calibration]:::amber
    H --> I[Failure Analysis Pipeline]:::rose
    I --> J[Grad-CAM Heatmap Interpretability]:::violet

Warning

Failure Analysis Spotlight: Under Blur Severity Level 4, accuracy drops to 11.80% (a Δ -61.87 pp loss from clean baseline). Under Resolution Severity Level 5, ECE spikes to 0.8520, highlighting severe overconfidence under resolution degradation.

Tech Stack: PyTorch · Torchvision · PathMNIST · Scikit-Learn · PyTest · Matplotlib
Repository Link: BioVision-Path GitHub Repository
Hugging Face Demo: BioVision-Path Space

`04` — BIOVISION-PATH

BIOMEDICAL COMPUTER VISION PIPELINE |

A multi-task biomedical computer vision pipeline covering classification, semantic segmentation, object detection, interpretability, and interactive inference.

`90.54% mAP@0.50`

_{YOLOv8 Object Detection}

`49.52% mAP@0.50`

_{Faster R-CNN Detection Baseline}

flowchart TD
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef dark fill:#1E293B,stroke:#334155,color:#F8FAFC,stroke-width:1px;

    A[Biomedical Histopathology Data]:::dark

    A --> B[Classification Task]:::blue
    A --> C[Segmentation Task]:::blue
    A --> D[Object Detection Task]:::cyan

    B --> B1[ResNet-18]:::blue
    C --> C1[U-Net]:::blue
    D --> D1[YOLOv8 — 90.54% mAP]:::cyan
    D --> D2[Faster R-CNN — 49.52% mAP]:::cyan

    B1 & C1 & D1 & D2 --> E[Grad-CAM / Interpretability]:::violet
    E --> F[Interactive Inference UI]:::violet

Tech Stack: Python · PyTorch · YOLOv8 · U-Net · Faster R-CNN · OpenCV · Grad-CAM · Gradio
Repository Link: BioVision-Path Repository

↑ HOME DOCK

03 — RESEARCH

IEEE Xplore Publication

Title: Explainable AI for Suicide Ideation Detection in Social Media Text

Identity Accent:

Core Focus: Research exploring explainable NLP approaches using deep learning architectures and feature attribution methods for early risk identification in social text streams.
Evaluated Architectures: BERTimbau · DistilBERT · XLM-RoBERTa · CNN-BiLSTM
Explainability Frameworks: Integrated Gradients · SHAP (SHapley Additive exPlanations)
Performance Metrics: Achieved 72.31% weighted F1 and 69.90% macro F1 across multi-class risk categories.

↑ HOME DOCK

04 — ENGINEERING STACK

[Models]      Python · PyTorch · OpenCV · Gemini API
[NLP / Vision] BERT · DistilBERT · XLM-RoBERTa · YOLOv8 · U-Net
[XAI & Eval]  SHAP · Integrated Gradients · PathMNIST · Scikit-Learn

[Frontend]    TypeScript · JavaScript · React 19 · Next.js 15
[Backend]     FastAPI · Express.js · Node.js · REST APIs
[Databases]   PostgreSQL · Prisma ORM · Drizzle ORM · Supabase · MongoDB

[Infrastructure] PostgreSQL · Redis · BullMQ Queue · Docker
[DevOps / Cloud] Git · Linux · Vercel · Railway · AWS S3
[Signal Processing] SciPy · NumPy · 4,096-point FFT · ISO 10816

[Unit / Integration] Vitest (170/170) · PyTest (39/39)
[API Verification]   Supertest / API Verification Tests (11/11)
[Validation Standards] ISO 10816 Vibration Standard · ECE Calibration

↑ HOME DOCK

05 — OTHER SYSTEMS

`CODEORIGIN` * Codebase intelligence & AST code parsing system.	`CAMPUSBUDDY` * Automated computer vision attendance & tracking platform.	`SIH PLATFORM` * National hackathon evaluation & grading infrastructure.
`CONTEST HOSTER` * Secure code execution container sandboxing system.	`REMOTE TREATMENT` * Intelligent patient telemetry & monitoring workflow.	`IMAGE SHARPENING` * Hardware-accelerated DSP image restoration pipeline.

06 — VERIFIED ENGINEERING

`170 / 170`

_{38 Test Files Executed}

`39 / 39`

_{BioRobust Benchmark}

`11 / 11`

_{SIH Platform Suite}

`01`

_{Published Research}

I don't just build systems. I validate them.

↑ HOME DOCK

07 — HOW I BUILD

flowchart LR
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef emerald fill:#10B981,stroke:#059669,color:#FFFFFF,stroke-width:2px;
    classDef indigo fill:#6366F1,stroke:#4F46E5,color:#FFFFFF,stroke-width:2px;
    classDef amber fill:#F59E0B,stroke:#D97706,color:#FFFFFF,stroke-width:2px;
    classDef rose fill:#F43F5E,stroke:#E11D48,color:#FFFFFF,stroke-width:2px;

    A[Research]:::violet --> B[Experiment]:::cyan
    B --> C[Engineer]:::blue
    C --> D[Validate]:::emerald
    D --> E[Deploy]:::indigo
    E --> F[Measure]:::amber
    F --> G[Iterate]:::rose

    G -.-> A

I focus on the engineering layer between AI research and usable software.

Models → Data → APIs → Backend → Infrastructure → Interfaces → Testing → Production

08 — ENGINEERING PRINCIPLES

`01` BUILD BEYOND THE MODEL

A model is only one component of an AI product. Data, APIs, queues, databases, interfaces, and infrastructure determine whether it becomes useful software.

`02` MEASURE EVERYTHING

Accuracy. F1. Calibration. Throughput. Latency. Failure modes. Testing. Evaluate under corruption, measure distribution shift, and back up assertions with test suites.

`03` DESIGN FOR REAL SOFTWARE

AI capabilities should become reliable systems people can actually use. Write modular, testable, type-safe code that scales gracefully.

CURRENT FOCUS

flowchart TD
    classDef violet fill:#8B5CF6,stroke:#7C3AED,color:#FFFFFF,stroke-width:2px;
    classDef cyan fill:#06B6D4,stroke:#0891B2,color:#FFFFFF,stroke-width:2px;
    classDef blue fill:#3B82F6,stroke:#2563EB,color:#FFFFFF,stroke-width:2px;
    classDef emerald fill:#10B981,stroke:#059669,color:#FFFFFF,stroke-width:2px;

    subgraph AI["AI SYSTEMS"]
        A1[LLM Applications]:::violet
        A2[Computer Vision]:::cyan
    end

    subgraph SOFTWARE["PRODUCTION SOFTWARE"]
        B1[APIs]:::blue
        B2[Data Pipelines]:::blue
        B3[Async Infrastructure]:::blue
    end

    subgraph PRODUCTS["USER-FACING PRODUCTS"]
        C1[Hiring Infrastructure]:::emerald
        C2[Industrial Telemetry]:::emerald
        C3[Biomedical Tools]:::emerald
    end

    AI --> SOFTWARE
    SOFTWARE --> PRODUCTS

09 — CONNECT

BUILDING SOMETHING INTERESTING?

Let's collaborate on AI systems, scalable backend infrastructure, or full-stack software applications.

SHAIK RAMEEZ BASHA

AI Systems Engineer — Turning research into intelligent software systems.

↑ HOME DOCK

Basharameez

SHAIK RAMEEZ BASHA

AI SYSTEMS ENGINEER

HOME — AI ENGINEERING SYSTEM

FEATURED SYSTEMS TILES

170 / 170

39 / 39

11 / 11

01

02 — SELECTED BUILDS

01 — APTIVUE

170 / 170

38

02 — ROTORDYN

4,096-Line

RMS

BPFO / BPFI

ISO 10816

03 — BIOROBUST

7,180

35

73.66%

0.0782

39 / 39

04 — BIOVISION-PATH

90.54% mAP@0.50

49.52% mAP@0.50

03 — RESEARCH

IEEE Xplore Publication

04 — ENGINEERING STACK

05 — OTHER SYSTEMS

CODEORIGIN

CAMPUSBUDDY

SIH PLATFORM

CONTEST HOSTER

REMOTE TREATMENT

IMAGE SHARPENING

06 — VERIFIED ENGINEERING

170 / 170

39 / 39

11 / 11

01

07 — HOW I BUILD

08 — ENGINEERING PRINCIPLES

01 BUILD BEYOND THE MODEL

02 MEASURE EVERYTHING

03 DESIGN FOR REAL SOFTWARE

CURRENT FOCUS

09 — CONNECT

BUILDING SOMETHING INTERESTING?

SHAIK RAMEEZ BASHA

Pinned Loading

Uh oh!

`170 / 170`

`39 / 39`

`11 / 11`

`01`

`01` — APTIVUE

`170 / 170`

`38`

`02` — ROTORDYN

`4,096-Line`

`RMS`

`BPFO / BPFI`

`ISO 10816`

`03` — BIOROBUST

`7,180`

`35`

`73.66%`

`0.0782`

`39 / 39`

`04` — BIOVISION-PATH

`90.54% mAP@0.50`

`49.52% mAP@0.50`

`CODEORIGIN`

`CAMPUSBUDDY`

`SIH PLATFORM`

`CONTEST HOSTER`

`REMOTE TREATMENT`

`IMAGE SHARPENING`

`170 / 170`

`39 / 39`

`11 / 11`

`01`

`01` BUILD BEYOND THE MODEL

`02` MEASURE EVERYTHING

`03` DESIGN FOR REAL SOFTWARE