# Raj Chhapariya — Complete Engineering Context (llms-full.txt) > Technical context for Raj Chhapariya, an AI / Data Engineer based in Bengaluru, India. > Figures were checked against each project repository; small samples are labelled as such. --- ## 1. Identity & Overview - **Full Name**: Raj Chhapariya - **Professional Role**: AI / Data Engineer - **Location**: Bengaluru, India (UTC+5:30) - **Canonical Website**: https://rajchhapariya.space - **Contact Email**: rajchhapariya8@gmail.com - **Verified GitHub**: https://github.com/Rajchhapariya - **Verified LinkedIn**: https://www.linkedin.com/in/rajchhapariya/ - **Verified Twitter / X**: https://x.com/raj_chhapariya - **Availability**: Open to projects, collaborations and AI / Data Engineering roles. ### Professional Biography Raj Chhapariya is an AI / Data Engineer based in Bengaluru, India, focused on building reliable data pipelines, in-process analytical engines, and evaluated LLM systems. His background spans Python, SQL, JavaScript, and TypeScript, with practical experience designing deterministic query guardrails and assessing LLM outputs for groundedness and citation faithfulness. Full-stack and backend development serve as supporting capabilities that allow him to take data systems from concept to production-ready interfaces. ### Core Operating Principles 1. **Accuracy and Grounding**: AI and analytical tools must verify outputs against underlying data and provide transparent citations. 2. **Safety by Architecture**: Restrict execution privileges (such as limiting SQL to single SELECT or WITH statements) rather than relying on unconstrained code execution. 3. **Clarity and Maintainability**: Write clear, modular code with structured Pydantic and TypeScript validation contracts. --- ## 2. Education & Credentials ### Education - **Degree**: Bachelor of Technology (B.Tech) - **Major**: Information Technology - **Institution**: Dr. A.P.J. Abdul Kalam Technical University - **Timeline**: 2020 – 2024 ### Certifications - **Google Data Analytics Professional Certificate**: Coursera (Issued May 2023) - **Cloud Computing Certification**: Honeywell Student Empowerment Program, Grade A (Issued Oct 2023) --- ## 3. Technical Competencies - **Core Data & Machine Learning**: Python, SQL, DuckDB, Pandas, NumPy, Scikit-learn - **AI & LLM Systems**: RAG Systems, OpenAI API, Rank-BM25, Pydantic, Prompt & Output Evaluation, Reciprocal Rank Fusion (RRF) - **Databases & Storage**: PostgreSQL, Supabase, MongoDB, Redis, REST APIs, Firebase - **Supporting Full-Stack**: TypeScript, JavaScript, Next.js (App Router, Server Components), React, Node.js, Streamlit - **Tools & Analytics**: Git, GitHub, Plotly, Power BI, Matplotlib, Seaborn - **Languages & Mobile**: Python, SQL, TypeScript, Java, Kotlin, Android Studio --- ## 4. Project Case Studies Figures below were checked against each project's repository. Where a figure comes from a small sample, the sample size is stated. ### 4.1. Agentic RAG Knowledge Assistant - **URL**: https://rajchhapariya.space/projects/agentic-rag-assistant - **Repository**: https://github.com/Rajchhapariya/Agentic-RAG-Knowledge-Assistant - **Status**: Open Source - **Role**: AI / Data Engineer (Aug 2026 – Present) - **Tech Stack**: Python, OpenAI API, Rank-BM25, NumPy, Pydantic, Streamlit - **What it is**: A retrieval-augmented generation pipeline over 10 research papers that plans queries, retrieves with BM25 and dense vectors fused by Reciprocal Rank Fusion (k = 60), audits the evidence, retries up to twice, then answers with citations, gives a partial answer, or refuses. - **Held-out results** (the dataset has 40 questions: 8 development and 32 held-out; results are from the 32-question held-out test set of 24 answerable, 4 unanswerable and 4 false-premise questions, and answer accuracy is calculated over the 24 answerable questions; frozen results from before later auditor fixes): - Hallucinations on unanswerable questions: 0 / 4 (the hybrid RAG baseline: 4 / 4). - Answer accuracy: 45.8% on answerable questions (the hybrid baseline: 60.4%). - Wrongly refused: 12.5% (3 of 24). - Average latency: 17.69 s against 1.86 s for the baseline. - Citation precision 55.2%, evidence recall 52.1%. - Tests: 90 test functions, 87 run offline by default. - **Limitations**: slow audit loop; conservative gate lowers accuracy; very small test sample; one false-premise question was answered instead of challenged, with a fix added after the evaluation run. ### 4.2. AI Data Analyst Agent - **URL**: https://rajchhapariya.space/projects/ai-data-analyst - **Repository**: https://github.com/Rajchhapariya/AI-Data-Analyst-Agent - **Status**: Open Source - **Role**: AI / Data Engineer (Aug 2026 – Present) - **Tech Stack**: Python, DuckDB, Plotly, Streamlit, Pydantic, Pandas - **What it is**: An agent that routes each question to one of four fixed tools (query_data, plot_chart, summary_stats, clarify). SQL is restricted to single SELECT/WITH statements using regex-based keyword guardrails (35 blocked keywords) on an in-memory DuckDB connection. This is a keyword blacklist, not a parser, and it is not engine-enforced read-only access or a guarantee that external file reads are blocked. Results are capped at 500 rows, and a numeric guard checks numbers in the answer against tool output. - **Results** (20-question benchmark on the bundled Superstore sales data): tool selection 20 / 20, execution success 20 / 20, answer correctness 17 / 20 (85%), numerical faithfulness 16 / 20 (80%), average latency 4,142 ms. The repository also lists "industry baseline" figures that are not measured comparisons. - **Limitations**: keyword-blacklist SQL guard (not engine-enforced read-only access); LLM routing can vary between runs; one in-memory table per session; single-step execution only; the faithfulness check covers numbers only. ### 4.3. Resume Roaster - **URL**: https://rajchhapariya.space/projects/resume-roaster - **Live Application**: https://www.atsroast.com/ - **Status**: Site live, with public scan and pricing pages (private repository); the features below are implemented in the codebase and some are gated by paid plans; customer usage is not published - **Role**: Founder & Independent Developer (Jul 2026 – Present) - **Tech Stack**: Next.js 16, TypeScript, Supabase (PostgreSQL with Row Level Security), OpenAI API, Tailwind CSS, Razorpay - **What it is**: A resume analysis and rewriting product. Implemented in the codebase: ATS-style parsing checks, a bullet rewriter that is prompted to avoid repeating action verbs (a prompt instruction, not a code check), summary, cover-letter and mock-interview generation, and PDF and Word export with publish-to-web. Razorpay checkout with server-side verification, webhooks, invoices, coupons and credits; Supabase accounts; Sentry; three Vercel cron jobs; Jest and Playwright setups. - **Not published**: usage figures and accuracy evaluations. ### 4.4. JanDarshan - **URL**: https://rajchhapariya.space/projects/satta-darshan - **Live Site**: https://jan-darshan.vercel.app - **Repository**: https://github.com/Rajchhapariya/JanDarshan - **Status**: Open Source, live - **Role**: Data & Full-Stack Engineer (Apr 2026 – Jul 2026) - **Tech Stack**: Next.js, TypeScript, MongoDB, Mongoose, Three.js, Tailwind CSS - **What it is**: An independent civic information platform for India. Records are curated from public sources (sansad.in, Election Commission of India, state portals) and stored with source, date and verification fields. Features include a 3D view of the Lok Sabha (543 seats) and Rajya Sabha (245 seats) and an interactive map of 36 states and UTs. - **Limitations**: data is curated by hand and can go out of date; it is not an official government portal. --- ## 5. Other Projects - **Truly** (Founder & Developer, Apr 2026 – Present): A social platform aimed at young urban Indians (ages 18–30), built around structured 60-minute video circles of eight people and one host. Seat capacity is enforced with an atomic conditional update inside a database transaction, with a unique index against duplicate bookings and an outbox for notifications. Live at https://trulyapp.online/. - **MandateStack** (Proprietary, Jul 2026): B2B SaaS subscription recurring billing platform for Indian businesses with Razorpay API, UPI AutoPay mandate integration, and automated dunning/retry webhooks. - **VyaparStack** (Production, Jul 2026): Statutory GST compliance and invoicing web application with automated CGST/SGST/IGST calculation across 36 states/UTs, in-memory pdf-lib invoice generation, and Neon Serverless PostgreSQL. Live: https://vyaparstack-nine.vercel.app/ - **Namma Metro Guide** (Android, published on Google Play, built solo): An unofficial Bengaluru Metro guide with route search and interchange details, fare estimates, station information, an interactive map, favourites and offline use. Kotlin, Jetpack Compose, Room, Maps Compose, Firebase. https://play.google.com/store/apps/details?id=com.rajch.nammametroguide --- ## 6. Technical Writing Five articles, reviewed for factual accuracy on 2026-11-05. Code samples are illustrative and nothing in them was benchmarked; statements about the author's own projects were checked against those repositories. - **Hybrid Retrieval Systems in Production** (https://rajchhapariya.space/blog/hybrid-retrieval-bm25-dense-embeddings-rrf): BM25, dense retrieval and Reciprocal Rank Fusion, with a worked RRF example. - **Deterministic Guardrails for LLM Agents** (https://rajchhapariya.space/blog/deterministic-guardrails-sql-pydantic-llm-agents): fixed tools, SQL keyword filtering versus AST parsing, and the limits of each. - **In-Process Columnar OLAP with DuckDB** (https://rajchhapariya.space/blog/in-process-columnar-olap-duckdb-architecture): columnar storage, vectorized execution and when an in-process engine fits. - **Next.js 16 App Router Architecture** (https://rajchhapariya.space/blog/nextjs-16-app-router-server-components-ssr): Server and Client Components, serialization, streaming and static regeneration. - **Evaluating Hallucination and Citation Faithfulness in RAG** (https://rajchhapariya.space/blog/evaluating-hallucination-citation-faithfulness-rag): evaluation metrics, a citation-structure check and offline testing. --- ## 7. Site Hierarchy & Navigation - **Home**: https://rajchhapariya.space/ - **About & Philosophy**: https://rajchhapariya.space/about - **Selected Work**: https://rajchhapariya.space/projects - **Resume & CV**: https://rajchhapariya.space/resume - **Download Resume PDF**: https://rajchhapariya.space/resume.pdf - **Privacy Policy**: https://rajchhapariya.space/privacy - **Sitemap XML**: https://rajchhapariya.space/sitemap.xml - **Robots Directives**: https://rajchhapariya.space/robots.txt - **LLM Summary Index**: https://rajchhapariya.space/llms.txt - **LLM Full Context**: https://rajchhapariya.space/llms-full.txt