Sai Rajesh
Tanikonda
I build data systems that improve inventory planning, supplier performance, and operational decision-making across supply chain and analytics workflows.
About Me
I am a data engineer with experience building scalable data pipelines and analytics systems. My work focuses on supply chain operations, including inventory planning, procurement analytics, and supplier performance tracking, enabling data-driven decision-making and operational efficiency.
I've built pipelines and analytics systems across supply chain and operational workflows, improving data latency, reliability, and decision speed for inventory, procurement, and supplier performance use cases.
My work spans batch and streaming architectures, analytics engineering, and data platform design, with a focus on supply chain visibility and operational performance.
Projects
SkyStream
Real-time global flight tracking pipeline. Ingests ADS-B telemetry from ~9,000 aircraft every 10 seconds via Kafka to Spark Structured Streaming to Redis, rendered on a GPU-accelerated deck.gl map with sub-5-second end-to-end latency. Includes email landing alerts and 48-hour historical trail queries via TimescaleDB. Demonstrates real-time system visibility and low-latency operational monitoring at scale.
FinSight
AI-assisted financial analytics platform that brings market dashboards, portfolio intelligence, research notes, curated news, and Telegram delivery into one full-stack product powered by FastAPI, React, DuckDB, Spark, Kafka, and dbt. Enables near real-time financial analysis and decision-making workflows by combining streaming data pipelines with scalable analytics and intuitive querying.
M5 Demand Forecasting & Inventory Replenishment
End-to-end ML pipeline built on Walmart's M5 dataset. Compares LightGBM gradient boosting against moving-average and seasonal-naive baselines, simulates reorder points with configurable safety-stock calculations, and delivers results through a three-persona Streamlit dashboard covering executive KPIs, model comparison, and planner-level replenishment recommendations.
Scalable URL Shortener
Cache-first URL shortener engineered for high read throughput. Redis handles the redirect hot path; async background workers flush click counters to PostgreSQL in batches. Real-time analytics UI supports operational monitoring and decision-making workflows. One-command Docker Compose deploy.
NYPD Arrests Pipeline
End-to-end analytics pipeline over 500K+ NYPD arrest records (2023-2024). Cleaned with Alteryx, modeled dimensionally, loaded to BigQuery for sub-second aggregations, visualized in an interactive Power BI dashboard by borough, offense type, and demographics. Transforms raw public data into structured datasets and dashboards for analyzing trends across geography and time.
EV Adoption Analysis
Visual deep-dive into U.S. electric vehicle adoption from 2010-2024. Analyzes charging infrastructure gaps, state-by-state adoption rates, and federal policy impact from 200K EVs in 2013 to over 4 million by 2024.
DIY CNC Plotter
2-axis CNC plotting machine built with Arduino UNO + GRBL firmware, 3D-printed components, and NEMA 17 stepper motors. Extended with an ESP32 IoT layer for wireless G-code transmission, browser-based control, and real-time telemetry.
Experience
- >Built and scaled a centralized analytics layer integrating multiple operational data sources, enabling real-time visibility into inventory movement, demand patterns, and material distribution across operations.
- >Built a unified analytics layer across 6 data sources, eliminating 20+ hours per week of manual reconciliation and improving data consistency.
- >Implemented data validation and reconciliation logic across inventory and distribution datasets, reducing discrepancies by 40% and improving accuracy of operational reporting.
- >Developed KPI frameworks to track inventory flows, resource utilization, and demand patterns, standardizing reporting across supply chain workflows.
- >Enabled stakeholders to identify inefficiencies in material allocation and improve planning, replenishment decisions, and overall operational efficiency.
- >Supported coursework in SQL and analytics, helping students and faculty apply data-driven approaches to problem-solving.
- >Assisted 140+ students in SQL and analytics concepts, improving understanding of data analysis and reporting.
- >Built dashboards to help faculty identify at-risk students and improve course planning and outcomes.
- >Worked across multiple client engagements in supply chain and retail domains, building data pipelines and analytics systems to support demand planning, inventory tracking, and operational decision-making.
- >Built supply chain data pipelines enabling near real-time inventory visibility and faster operational response across warehouse and distribution systems.
- >Designed dimensional data models for inventory, supplier performance, and shipment analytics, improving reporting consistency and planning accuracy.
- >Analyzed supplier performance, lead times, and demand trends to support better purchasing and replenishment decisions.
- >Reduced data latency from 12+ hours to under 1 hour using optimized batch and streaming pipelines.
- >Improved query performance and reduced compute costs through efficient data modeling and transformation strategies.
Blog Posts
Deep dives into the systems I build, the architecture, the decisions, and what I learned.
Every layer of the system: Kafka checkpoint gotchas that blanked the map, why Redis writes must precede Postgres, how deck.gl handles 9,000+ GPU-accelerated points, and the four production bugs I actually hit.
Read PostHow I combined market data pipelines, warehouse modeling, portfolio tooling, news aggregation, and a natural-language query layer into one end-to-end investing product.
Read PostHow LightGBM, leakage-safe feature engineering, and a safety-stock simulation combine to turn Walmart's M5 dataset into actionable replenishment recommendations, with a three-persona Streamlit dashboard at the end.
Read PostHow cache-first architecture and async click tracking combine to build a URL shortener that stays fast under heavy read load without sacrificing analytics accuracy.
Read PostProfiling, cleaning, and modeling 500K+ arrest records into an interactive Power BI dashboard that surfaces borough-level crime patterns and demographic trends.
Read PostFrom 200K EVs in 2013 to over 4 million by 2024, a state-by-state breakdown of adoption rates, charging infrastructure gaps, and the policy drivers behind the transition.
Read PostContact
I'm actively looking for data engineering roles where I can work on high-throughput streaming systems, large-scale data infrastructure, and the tooling that makes data reliable at scale.