Immediate JoinerWelcome to my portfolio

Hi, I'm Harshwardhan

Building production-grade backend and AI-powered applications with Python, FastAPI, REST APIs, and SQL — from LLM-powered document intelligence and advanced RAG to semantic search and Text2SQL systems that solve real business problems.

Get In TouchDownload CV
Find me on
Harshwardhan Gullapalli

1.5+ Years

Experience

RAG + LLM

AI-Focused Backend

Scroll Down
Python•FastAPI•SQL•REST APIs•Pydantic•Async Programming•LangGraph•RAG•Hybrid Search•BM25•RRF•Reranking•Qdrant•Pinecone•Semantic Search•PostgreSQL•MySQL•Redis•Docker•Docker Compose•AWS ECS•EFS•ALB•CloudFormation•Pytest•Ragas•Mypy•Python•FastAPI•SQL•REST APIs•Pydantic•Async Programming•LangGraph•RAG•Hybrid Search•BM25•RRF•Reranking•Qdrant•Pinecone•Semantic Search•PostgreSQL•MySQL•Redis•Docker•Docker Compose•AWS ECS•EFS•ALB•CloudFormation•Pytest•Ragas•Mypy•

About Me

Where backend engineering meets AI

Harshwardhan Gullapalli

1.5+ Years

Software Engineering

MCA (AI & DS)

D. Y. Patil University

Backend + GenAI

FastAPI, RAG, LangGraph

Pune, India

Immediate Joiner

I'm a Software Engineer with 1.5 years of experience building production-grade backend and AI-powered applications using Python, FastAPI, REST APIs, and SQL. With an MCA in Artificial Intelligence & Data Science from D. Y. Patil International University, I specialize in shipping AI capabilities into real-world business workflows.

My focus spans LLM-powered document intelligence, Retrieval-Augmented Generation (RAG), semantic search, and automated data classification — from parsing tax-facing documents into structured outputs to building enterprise copilots over knowledge bases.

I bring a strong foundation in backend architecture, asynchronous programming, and API development, and I love combining AI with deterministic business logic to build reliable, scalable systems.

Backend Engineering

Scalable FastAPI services, REST APIs, async programming, and Pydantic-validated data models for production workloads.

GenAI & LLM Systems

LLM integration with structured outputs, tool calling, and LangGraph orchestration layered over deterministic business logic.

RAG & Semantic Search

Hybrid search with BM25, reciprocal rank fusion, and cross-encoder reranking on Qdrant and Pinecone.

Cloud & DevOps

Containerized deployments with Docker, Docker Compose, and AWS ECS, EFS, ALB, and CloudFormation.

Experience

Professional Journey

Full Stack Developer

Radiant Services

Solapur, India

September 2024 - March 2026Full-time

Development focused on AI-powered tools for Chartered Accountants — backend services, LLM-powered document processing, RAG, and automated accounting workflows.

Key Contributions
  • Developed backend services for a UAE Corporate Tax Filing platform using Python, FastAPI, MySQL, REST APIs, and asynchronous programming, automating key workflows for Chartered Accountants (CAs).
  • Built an LLM-powered document processing pipeline that extracted financial and tax information from uploaded documents and generated structured outputs using Pydantic and TypedDict, enabling automated pre-population of tax-related fields.
  • Implemented deterministic financial and tax-calculation logic aligned with the filing workflow, allowing CAs to modify input variables and iteratively evaluate tax liability before final submission.
  • Developed an LLM-based Trial Balance auto-mapping system using predefined domain rules to classify accounting particulars into Account Head, P&L/Balance Sheet, Group, and Sub-Group, reducing manual classification time by approximately 75%.
  • Designed and integrated a private Retrieval-Augmented Generation (RAG) system using Pinecone over confidential tax laws, regulations, and financial documents, enabling contextual retrieval and question answering for tax-related queries.
PythonFastAPIMySQLREST APIsAsync ProgrammingPydanticLLMRAGPinecone

Projects

What I've built

From enterprise AI copilots to production backend systems and automated accounting workflows.

Enterprise SRE Copilot

Advanced RAG & Text2SQL Platform

An enterprise-grade conversational copilot that intelligently routes natural-language queries across RAG, Text2SQL, and hybrid workflows using a stateful LangGraph orchestration layer.

  • Advanced retrieval pipeline combining dense + sparse hybrid search, Reciprocal Rank Fusion (RRF), cross-encoder reranking, HyDE, CRAG with web fallback, and Self-RAG reflection/regeneration over an intentionally noisy knowledge base.
  • Text2SQL with human-in-the-loop approval using LangGraph interrupts, PostgreSQL schema-aware SQL generation, SELECT-only enforcement, and validation before query execution.
  • 5-tier Redis caching architecture for embeddings, intent routing, SQL generation, SQL results, and RAG answers — with workload-specific TTLs and cache observability.
  • 9-layer security and guardrail pipeline: input validation, prompt-injection detection, JWT authentication, rate limiting, token budgets, PII redaction, content moderation, spotlighting, and structured output validation with retry mechanisms.
FastAPILangGraphHybrid SearchRRFRerankingHyDECRAGSelf-RAGText2SQLRedisDockerAWS ECS/EFS/ALB/CFNPytestRagas
Key Capabilities

Hybrid Retrieval

Dense + sparse search, RRF fusion, reranking, HyDE, CRAG & Self-RAG

5-Tier Redis Cache

Embeddings, routing, SQL generation, results & answers with TTLs

9-Layer Guardrails

Input validation, injection detection, auth, rate limiting, PII redaction

LangGraph Stateful

Interrupts, human-in-the-loop approval & multi-agent orchestration

UAE Corporate Tax Filing Platform

Work Project — Backend for Chartered Accountants

Backend services for a UAE Corporate Tax Filing platform that automates key workflows for Chartered Accountants.

  • FastAPI + MySQL backend with REST APIs and async programming to orchestrate the full filing workflow.
  • LLM-powered document processing pipeline extracting financial and tax information into structured outputs via Pydantic and TypedDict, pre-populating tax fields.
  • Deterministic financial and tax-calculation logic letting CAs adjust inputs and iteratively evaluate tax liability before submission.
PythonFastAPIMySQLREST APIsAsyncPydanticLLM
Client / Work project — source not public

LLM Trial Balance Auto-Mapping

Work Project — Automated Data Classification

An LLM-based system that classifies accounting particulars into the right financial statement categories using predefined domain rules.

  • Classifies particulars into Account Head, P&L/Balance Sheet, Group, and Sub-Group with deterministic domain rules driving LLM reasoning.
  • Reduced manual classification time by approximately 75% for day-to-day CA workflows.
  • Private RAG over confidential tax laws and regulations using Pinecone for contextual retrieval and Q&A.
LLMRAGPineconeRule-based AIPydanticPrompt Engineering
Client / Work project — source not public

Skills

Technologies I work with

Languages

Python95%
SQL85%

Backend

FastAPI90%
REST APIs90%
Async Programming85%
Pydantic88%

GenAI / LLM

LLM Integration92%
Structured Outputs90%
RAG92%
LangGraph85%
Prompt Engineering88%
Tool Calling82%

Retrieval & Search

Qdrant82%
Pinecone85%
Hybrid Search86%
BM2584%
RRF82%
Reranking80%
Semantic Search88%

Databases & Cache

PostgreSQL88%
MySQL90%
Redis80%

Cloud & DevOps

Docker85%
Docker Compose82%
AWS ECS78%
EFS / ALB75%
CloudFormation72%

Testing & Quality

Pytest86%
Ragas78%
Mypy80%
Ruff82%
Git90%

Services

How I can help

AI-powered, production-grade software delivered end-to-end — from backend services to intelligent retrieval systems.

Backend Development

Production-grade FastAPI services, REST APIs, async workflows, and clean, typed data models with Pydantic.

  • Scalable API design & architecture
  • Async I/O and background processing
  • Schema validation & typed outputs

GenAI & LLM Solutions

LLM-powered features grounded in your business logic with structured outputs and agentic orchestration.

  • LLM integration & structured outputs
  • LangGraph stateful orchestration
  • Tool calling & human-in-the-loop flows

RAG & Semantic Search

Reliable retrieval over your own data with hybrid search, reranking, and evaluation-backed quality.

  • Hybrid search: BM25 + vectors, RRF
  • Cross-encoder reranking & HyDE
  • Qdrant / Pinecone vector stores

Cloud & DevOps

Containerized, observable, and deployable systems on AWS with repeatable infrastructure.

  • Docker & Docker Compose
  • AWS ECS, EFS, ALB, CloudFormation
  • Pytest, Ragas, Mypy, Ruff quality gates
Let's discuss your project

Education

Academic background

MCA (Artificial Intelligence & Data Science)

D. Y. Patil International University

Pune | 2022 - 2024

CGPA: 7.85
  • Specialized in AI, Machine Learning, Deep Learning, and Data Science
  • Hands-on projects spanning NLP, LLMs, and applied intelligent systems
  • Strong foundation for building production-grade AI applications

BCA (Bachelor of Computer Applications)

Hirachand Nemchand College of Commerce

Solapur | 2019 - 2022

CGPA: 8.3
  • Solid grounding in Computer Science fundamentals
  • Programming in Python, Java, and web technologies
  • Database management and software engineering principles

Contact

Let's work together

I'm an immediate joiner and open to Software Engineer & AI Engineer roles. Let's build something great together!

Email

gullapalli.harsh@gmail.com

Phone

+91 7447643226

Location

Pune, Maharashtra, India

LinkedIn

harsh-gullapalli

GitHub

harsh-gullapalli

Ask me?

Chat with AI