Software Engineer — Data & AI Systems

Software that creates real value for business.

I’m Anas. I build retail software, data pipelines, and applied AI.

About me

I’m Anas, a software engineer working across data systems, backend architecture, and applied AI. I care about clean, maintainable software that holds up when people depend on it.

My work ranges from a retail POS in daily production use to Azure analytics pipelines, computer vision, and real-time voice agents. I enjoy the practical trade-offs that make these systems fast, reliable, and useful.

Selected work

From daily operations to applied AI

Featured SystemIn Daily Production
Retail POS & Cloud Analytics Platform
A point-of-sale system built for daily retail operations, with fast cashier workflows and automated cloud backups. An Azure SQL pipeline turns store data into Power BI reporting.
1. Counter POSReact · Tailwind
2. Local engineFlask · SQLite
3. Cloud backupAzure Blob
4. WarehouseAzure SQL
5. AnalyticsPower BI
3–5×Checkout Speedupvs. previous setup
4 CashiersDaily ProductionCounter billing & stock
Explore architecture
Data Pipeline & System ArchitectureClick a node to inspect
Stage 1: Counter Terminal UIReact • Tailwind CSS • shadcn/ui
Designed for fast keyboard-driven cashier workflows. Features instant barcode lookup, responsive cart recalculations, discount rules, and sub-second receipt generation with zero screen stutter.
Stage 2: Embedded Transaction EngineFlask • Marshmallow • SQLite • Redis Queue
Runs on-premise at the retail store with SQLite for low-latency atomic transactions during peak rush hours. Marshmallow enforces strict schema validation, while Redis Queue handles async thermal printing and inventory ledger updates.
Stage 3: Automated Cloud Backup & ReplicationAzure Blob Storage • Azure Key Vault • Python CLI
An automated nightly process generates encrypted database snapshots and synchronizes them to versioned Azure Blob Storage containers, allowing recovery from the latest saved snapshot without manual cashier overhead.
Stage 4: Automated Data Warehouse & ETLAzure SQL • Azure Container Apps • Docker • OBT
A containerized pipeline pulls the latest blob backup, verifies integrity checksums, restores tables into Azure SQL, and transforms normalized relations into an optimized One Big Table (OBT) schema for high-speed analytical queries.
Stage 5: Executive Business IntelligencePower BI • DAX Measures • Automated Refresh
Feeds directly into Power BI dashboards to track daily revenue velocity, product gross margins, cashier transaction volume, inventory turnover, and stock-out forecasts across all store categories.
ReactTailwind CSSFlaskSQLiteRedis QueueAzure Blob StorageAzure SQLPower BIDocker
Real-time AI · LiveKit · Python
Voice AI customer support agent
A browser-based voice agent for food delivery support, with speech, language-model, and voice fallbacks to keep conversations running.
Computer vision · OpenCV · YOLO
End-to-end computer vision pipeline
Video processing, object detection, segmentation, custom dataset creation, and YOLO training — from raw footage to a working model.
Document AI · PaddleOCR · LLMs
Invoice data extraction pipeline
Pulls structured fields — invoice number, GST number, grand total, tax total — out of invoice images and PDFs using PaddleOCR with an LLM on top.

Work Experience

Summer InternIIIT HyderabadMar–May 2026
Summer internship at IIITH — applied research and engineering across computer vision, media processing pipelines, and data systems.
PythonOpenCVFFmpegYOLO

Writing

All essays →

Aug 14, 202616 min read
Azure Entra ID: Service Principals vs Managed Identities
Azure resources have no identity by default. The complete mental model of Entra ID, Service Principals, and Managed Identities built from first principles.

Education

B.Tech — Computer ScienceMuffakham Jah College of Engineering & Technology (MJCET), Hyderabad, India
2022–2026
Coursework across Operating Systems, Database Management Systems, Computer Networks, Data Structures, Artificial Intelligence, and Machine Learning.