Skip to content
View HarshitSharma14's full-sized avatar

Block or report HarshitSharma14

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
HarshitSharma14/README.md

LinkedIn LeetCode Codeforces Gmail


🧠 Technologies & Tools

πŸ’» Languages
βš™οΈ Frameworks
πŸ€– AI / Agentic
πŸ—„οΈ Data & Infra
☁️ AWS
🧩 Core CS

πŸ“Œ Pinned Projects

πŸ—οΈ Credo

SaaS CRM

Built during my internship β€”> a full CRM platform with org-level isolation.

πŸ”Ή Configurable roles, hierarchies & flexible permissions
πŸ”Ή Designed 65+ relational tables for complex workflows
πŸ”Ή Lead pipelines, follow-ups & audit logs
πŸ”Ή Role-based dashboards per organization

What I learned

  • Multi-tenant SaaS architecture
  • Large-scale relational database design
  • Building complex backend workflows
  • Vue + production frontend structure

βš–οΈ LegalOS

RAG AI

AI-powered legal research assistant for Indian law.

πŸ”Ή Ingests Indian Central & State Acts
πŸ”Ή Document cleaning, chunking & embedding
πŸ”Ή Vector search for contextual legal answers
πŸ”Ή Local SLM inference for privacy-friendly analysis

What I learned

  • Prompt engineering for structured responses
  • Designing and building RAG pipelines
  • Writing cleaner code through PR reviews
  • Working with embeddings and vector search

πŸš€ Featured Project

πŸ₯ Health Insurance Claims Processor

Multi-Agent FastAPI Tests

AI pipeline that adjudicates health insurance claims using multi-agent orchestration.

πŸ”Ή 4-agent pipeline: Eligibility β†’ Medical Review β†’ Financial Analysis β†’ Decision
πŸ”Ή 132 passing tests + 12/12 eval harness with full processing traces
πŸ”Ή Gemini API + FastAPI backend + Vite frontend, deployed on Render + Vercel
πŸ”Ή APPROVED / PARTIAL / REJECTED / MANUAL_REVIEW with confidence scores

What I built: Production-grade multi-agent system with automated evaluation harness


🌐 Open Source Contributions

βš–οΈ LegalOS β€” 20+ Pull Requests

RAG Contributor

Contributed to an AI-powered legal research assistant for Indian law.

πŸ”Ή 20+ merged PRs across RAG pipeline, chunking, and retrieval
πŸ”Ή Improved benchmark from 33.93% β†’ 43.37% accuracy
πŸ”Ή Vector search, local SLM inference, document preprocessing
πŸ”Ή Real collaborative open-source engineering

πŸ”¬ GSoC Research Contributions

Open Source Research

Contributed to OREL group's Google Summer of Code projects.

πŸ”Ή Scientific open-source tooling contributions
πŸ”Ή PRs to OREL-group/GSoC
πŸ”Ή Research-adjacent engineering work

πŸ“Š GitHub Stats


Curious about unique problems and how technology and AI can solve them.

Pinned Loading

  1. Credo-DSA-CRM Credo-DSA-CRM Public

    Multi-tenant CRM built solo in 40 days β€” 65+ relational tables, RBAC, audit logs, SSR (SiteGuru internship)

    1

  2. prasenjitb-cloud/legalos prasenjitb-cloud/legalos Public

    Python 1