Engineering Projects

Apr 2026 – May 2026

Developed a high-performance RAG-driven application tailored for context-aware technical document synthesis and context retrieval execution. Utilized advanced segmentation logic to split structural data assets, converting unstructured textbook formats into persistent high-dimensional embeddings optimized for exact matches.

• Engineered MMR-based (Maximal Marginal Relevance) retrieval structures inside ChromaDB vectors to reduce token generation hallucination metrics.
• Integrated FastAPI microservice endpoints connecting Mistral AI modules to scale processing execution rates.

Mar 2026 – Apr 2026

Designed an end-to-end regression modeling service constructed to compute ride-hailing operational fare predictions across multi-variant localized market constraints. Implemented rigorous feature engineering pipelines mapping structural weather dependencies, historical grid congestion indices, and spatio-temporal coordinates.

• Conducted comprehensive Exploratory Data Analysis (EDA) loops mapping out anomalies across dynamic localized marketplace models.
• Deployed low-latency lookup wrappers inside TensorFlow and FastAPI, binding operational states to real-time asynchronous databases.

Jan 2026 – Feb 2026

Analytics & Modeling Workflow

An end-to-end modern data analytics architecture engineered to ingestion endpoints. Designed a unified Bronze → Silver → Gold storage topology, automated ML-driven client segmentation, and constructed high-impact executive intelligence layouts.

• Built layered lakehouse data structures (Bronze, Silver, Gold) inside cloud repositories.
• Engineered custom behavior vectors & automated machine learning clustering segmentation models.
• Created comprehensive executive dashboards detailing cross-functional revenue patterns, retention loops, and risk margins.

Dec 2025 – Jan 2026

Data Transformation Infrastructure

An production-grade ELT (Extract, Load, Transform) engineering pipeline architected to handle massive retail inventory data. Transformed unorganized flat CSV staging layers into clear dimensional star schemas to accelerate optimization tracks.

• Configured cloud storage warehouses to ingest unstructured transactional datasets seamlessly.
• Modeled analytical structures cleanly by shifting processing rules over to robust dbt Core dependencies.
• Generated interactive Power BI operational visuals targeting granular high-frequency metrics.

Nov 2025 – Dec 2025

Designed and executed a campus-wide localized content routing and queuing infrastructure matching student print requests to nearby physical print terminals. Employs asynchronous cloud workers to process heavy data assets on the fly.

• Integrated Cloudinary transformation APIs to handle background image-to-PDF asset conversions securely.
• Utilized Firebase real-time event distribution hooks to synchronize shop-owner queue tracking dashboards instantly.