Muhammad Tauha Kashif

Academic

Career

Verbatim text

Muhammad Tauha Kashif
Cell: 923334727249 |  Email: muhammad.tauha@outlook.com
LinkedIn: https://www.linkedin.com/in/muhammad-tauha-
Address: House 25, Block 16 , Sector b-i, township , Lahore , Pakistan
PROFESSIONAL PROFILE
Fresh Software Engineering graduate passionate about data engineering, with hands-on experience in AWS and GCP pipelines,
real-time processing, and dashboarding. Focused on developing efficient data processing systems that transform raw data into
actionable insights. Experienced in batch data processing, cloud infrastructure optimization, and building analytics dashboards that
drive business decisions.
EDUCATION
Bachelors in Software Engineering
School of Electrical Engineering and Computer Sciences , Islamabad (2022)
INTERNSHIP EXPERIENCE
Software Productivity Strategists (SPS) Inc.
01-Jul-2025 - 22-Sep-2025
DevOps/DataOps Intern • Shipped CI/CD for containerized data services with Jenkins (multibranch) and reproducible Docker builds;
gated quality via pre-commit (Black, Ruff). • Instrumented ETL containers + hosts with Prometheus/Grafana (cAdvisor, Node
Exporter, Alertmanager); built scrape configs & dashboards for latency, errors, and capacity. • Standardized project scaffolding (.env,
.dockerignore, Makefile, READMEs) so new services go from repo ’ deploy with fewer steps and cleaner diffs.
Buildables
22-Jul-2025 - 18-Oct-2025
Data Engineering Intern • Built production-grade ETL pipelines with hash-based change detection (MD5) and PostgreSQL upserts,
processing incremental loads while maintaining full audit trails and execution metadata for data lineage tracking. • Designed star
schema data warehouse implementing SCD Type-2 for historical tracking; wrote complex analytical queries using CTEs, window
functions, and joins to derive business insights from e-commerce transaction data. • Optimized large dataset processing by
benchmarking Pandas vs Dask vs Polars on 2M+ records, achieving 10-15x performance gains through lazy evaluation and Parquet
columnar storage—reducing file sizes by 75%. • Developed modular data quality framework with custom cleaners for standardizing
inconsistent formats (dates, currencies, names), achieving 99.7% parse success across messy real-world datasets. • Containerized
entire data stack using Docker Compose with isolated PostgreSQL instances, automated schema initialization, and health checks—
ensuring reproducible deployments across environments.
FINAL YEAR PROJECT
AI Development Environment Troubleshooting Copilot
- Autonomous AI Agent System: Developed an intelligent troubleshooting copilot that automates diagnosis and resolution of
development environment configuration issues (Docker, package managers, CLI toolchains) using LLM-powered workflow
orchestration. - System Profiling & Context Extraction: Built modular CLI utilities for capturing structured error traces and
comprehensive system snapshots (hardware, processes, services, network, installed packages) to provide rich diagnostic context. -
Hybrid Web Architecture: Engineered full-stack solution with React/TypeScript frontend and FastAPI backend, featuring real-time
WebSocket event streaming, chat-based interface, and agent workflow visualization. - RAG-Enhanced Troubleshooting Pipeline:
Implemented vector-based semantic retrieval system using ChromaDB with e5-base-v2 embeddings for context-aware error
diagnosis, combined with LangGraph-based multi-stage workflow orchestration (initialization → context gathering → step generation
→ error resolution). - Structured Data Pipeline: Designed JSONL-based training data format for error normalization and context
requirement detection across multiple domains (Python, Node.js, Docker, Git), enabling future fine-tuning and knowledge base
expansion. - Production-Ready Features: Integrated error recovery mechanisms, command safety verification, diff-based state

AI enrichment