I design and build reliable, production-grade AI applications, document intelligence pipelines, and data platform infrastructure. I bridge the gap between machine learning capabilities and scalable software engineering.
- Languages: Python, TypeScript, SQL (PostgreSQL, MySQL), Bash
- AI & ML: LLM Prompt Engineering, RAG (pgvector, ChromaDB), PaddleOCR, Computer Vision (MediaPipe, OpenCV)
- Backend & APIs: FastAPI, Node.js (Express), REST APIs, WebSockets
- Frontend: Next.js (React), Zustand, TailwindCSS, Recharts
- Infrastructure & DevOps: Docker, Docker Compose, Nginx, Turborepo, pnpm Workspaces, CI/CD (GitHub Actions)
A natural-language business intelligence application that translates user queries into audited SQL and interactive charts.
- Stack: Python, FastAPI, SQLAlchemy 2, Alembic, React, TypeScript, Zustand, Recharts, Docker.
- Key Features: Implemented a model-backed prompt engine, safe SQL query validation schemas, database connector adapters (Postgres, DuckDB, MySQL), and a grid-based canvas.
An enterprise document processing and form automation platform with OCR and structured extraction.
- Stack: Next.js (TypeScript), TailwindCSS, FastAPI, Python, PaddleOCR, Docker.
- Key Features: Engineered a decoupled OCR microservice, structured layout extraction pipelines, and automated UI integration tests via Playwright.
Full-stack invoice extraction with local OCR, structured parsing, and deterministic offline fallbacks.
- Stack: Python, FastAPI, React, TypeScript, PaddleOCR, SQLite, Pydantic.
- Key Features: Combined OCR text extraction with Pydantic schema validation for deterministic extraction of items, vendor info, and total values.
An LLM-driven research, content-generation, and automated publishing system.
- Stack: JavaScript, Node.js, Puppeteer, Python, Gemini API.
- Key Features: Automated web scraping, dynamic infographic generation via HTML-to-image renderers, and automated multi-channel publishing scheduling.
Cache-first Node.js API for daily fuel-price data across Indian cities and states.
- Stack: JavaScript, Node.js, Express, Redis.
- Key Features: Implemented cache-first retrieval logic, automated daily scraping of state fuel prices, and structured REST API endpoints.
- LinkedIn: aniket-thakur-a23a9b372
- Twitter: @07anni04