Rajesh Sharma
Full-Stack Engineer & Data Scientist
Bridging logic, design, and data to deliver impactful software.
Full-Stack Engineer & Data Scientist
Bridging logic, design, and data to deliver impactful software.
I build intelligent, scalable, and automation-first systems that power real-time decisions, optimize processes, and unlock data-driven growth. From architecting multi-tenant platforms and orchestrating ETL pipelines to deploying NLP and LLM-powered applications, I bring together software engineering, data science, and cloud expertise to turn complex problems into seamless digital products. My systems have served more than 10K users, improved engagement by over 20 percent, reduced manual workloads by 60 percent, and saved more than $20K per month in operational costs while driving measurable business growth and efficiency. Whether it’s accelerating research with ML, scaling nonprofit platforms, or automating customer intelligence : I engineer solutions that create measurable impact.
Design and build scalable, responsive applications using React, Django, and RESTful APIs to balance performance with maintainability.
Develop ETL pipelines, automate multi-source ingestion (APIs, web scraping, databases), and deliver clean, structured data for analytics and ML.
Build models for classification, anomaly detection, and topic extraction using LLMs, Transformers, and traditional ML, focusing on impact and interpretability.
Deploy and monitor cloud-native apps using AWS (EC2, S3, Lambda), Docker, and CI/CD pipelines — optimizing performance and cost.
Create interactive dashboards with drill-down metrics, executive summaries, and real-time insights to support informed decision-making.
Architect secure, token-authenticated, modular systems that enable multi-tenant SaaS platforms, scalable microservices, and efficient backend design.
From Ideas to Deployments
Built a production-grade full-stack application featuring custom RAG (Retrieval-Augmented Generation) from scratch to solve meal planning and food waste challenges. Engineered hybrid search combining BM25 keyword matching (60%) and vector similarity (40%) across 768-dimensional embeddings to retrieve relevant recipes with 95%+ accuracy. Integrated local LLM (Llama 3.2) via Ollama for privacy-first conversational AI with zero API costs. Developed intelligent algorithms for unit normalization using Pint library, ingredient consolidation across recipes, and shelf-life optimization that reduces food waste by 30% through graph-based meal sequencing.
Built an end-to-end data intelligence system to collect, process, and analyze 584K+ digital asset records for market research and policy impact studies. Developed resilient ETL pipelines using Python to extract 84K+ cryptocurrency transactions, plus 500K+ NFT records from OpenSea's marketplace API. Automated pipeline that achieved 95%+ data completeness while reducing API failures by 80%.Conducted comprehensive analysis of OpenSea's January 2022 policy changes, quantifying impacts on creator behavior, collection monetization, and secondary market dynamics to deliver actionable insights for fundraising strategies and platform economics research.
Built a Python library to automate OCR text extraction from HathiTrust's 17M+ book database using OAuth1 authentication and parallel processing. Designed fault-tolerant data extraction workflows with checkpointing, retry logic, and session management to handle multi-GB datasets efficiently. Implemented chunked file processing (200K rows/batch) and concurrent API requests (100 workers) with intelligent rate-limiting strategies, reducing text retrieval time by 70% across 10,000+ pages while ensuring 100% completion on interrupted workflows.
Built a bipartite graph-based system to recommend meal plans based on nutritional values and ingredient compatibility. Designed a custom graph embedding (not based on conventional DeepWalk) and applied K-Means clustering to group similar ingredients. Enabled allergy-aware substitutions for dietary constraints. The model supports ingredient-level personalization and aims to assist hospitals and individuals in preparing nutrition-optimized recipes with minimal human intervention.
Performed in-depth exploratory data analysis on Rent the Runway dataset, uncovering trends in customer behavior, product ratings, and fit satisfaction. Analyzed rental patterns across age groups, body types, and seasons. Applied sentiment analysis on reviews and visualized insights using 2D/3D plots to inform customer segmentation and occasion-based marketing strategies.
Designed a custom embedding approach for a bipartite ingredient-nutrient graph to represent nutritional relationships in lower-dimensional space. Built a biased random walk algorithm influenced by edge weights (normalized nutrition values) to better preserve feature significance during embedding. This technique improved ingredient clustering for dietary recommendation systems.
Built an NLP-powered pipeline to extract executive names and job titles (e.g., CEO, VP) from M&A press releases using web scraping, Named Entity Recognition, and relationship extraction. Automated the analysis of SEC filings and business blogs to identify key delegates involved in high-value deals, reducing manual effort in business intelligence workflows.
Developed a recommendation system using H&M’s customer, article, and transaction data (31M+ records) to generate personalized fashion suggestions based on purchase history and demographic clusters. Leveraged association rules, unsupervised clustering (by age), and metadata preprocessing to enhance the shopping experience and reduce decision fatigue. Optimized data handling using GPU-based processing (cudf) and memory-efficient formats (.parquet) for scalable insights.
Skilled in full-stack engineering, machine learning, data engineering, and cloud deployment. I bring a product-first mindset and engineering depth to solve business, research, and nonprofit challenges at scale.
MS in Computer Science - Indiana University Bloomington
Specialization Data Science & AI - IIIT Delhi
B.Tech in Computer Science - Guru Nanak Dev University
Let’s build something meaningful together. Whether it’s an idea, project, or collaboration, I’m just a message away.